/iemg/ & /pmpg/ - In-Ear Monitor & Portable Music Player General

Meme edition

梗图版。

How to request advice:

如何寻求帮助：

>Budget

>Intended use (media, source, environment)

>用途（媒体、音源、使用环境）

>Frequency response preference and music examples

>频响偏好及音乐示例

>Past gear and your thoughts on them

>用过的设备及你的评价

FAQ:

>Where do I buy IEMs?

>我在哪里买IEM？

Amazon, Aliexpress, Linsoul, Hifigo, Shenzhenaudio

亚马逊、速卖通、Linsoul、Hifigo、Shenzhenaudio

>Shopping Guide (IEMs, PMPs, Cables, Ear Tips, etc.):

>购物指南（IEM、PMP、线材、耳塞套等）：

https://consoomer-guide.pages.dev/

>/iemg/'s Journal:

>/iemg/ 的日志：

https://iemgazette.pages.dev/

>EQ Guide (EQ 101, Targets, Myths & Misconceptions, Papers, etc.):

>EQ指南（EQ入门、目标曲线、迷思与误解、论文等）：

https://iem-eq-guide.pages.dev/

>Measurements:

https://iem-eq-guide.pages.dev/measurement-databases/

>Budget Wire Over-Ear IEMs:

>预算款有线包耳式IEM：

• Tanchjim Bunny DSP (Mild U/5-Band PEQ) - $22

• Tanchjim Bunny DSP（轻微U型/5段PEQ） - $22

• Truthear Gate (Mild V) - $22

• Truthear Gate（轻微V型） - $22

• EPZ Q1 Pro (V-shape) - $35

• EPZ Q1 Pro（V型） - $35

>Bullet IEMs:

>子弹头IEM：

• Tanchjim Zero Ultima DSP (Mild U/5-Band PEQ) - $22

• Tanchjim Zero Ultima DSP（轻微U型/5段PEQ） - $22

• Sony IER-EX15C (Warm/Dark) - $30

• Sony IER-EX15C（暖/暗） - $30

• Etymotic ER2XR (Towards Neutral) - $140

• Etymotic ER2XR（偏中性） - $140

>Flathead Earbuds:

>平头耳塞：

• NICEHCK Vido (Warm) - $5

• NICEHCK Vido（暖） - $5

• Yincrow X6 (Warm) - $10

• Yincrow X6（暖） - $10

>USB-C DACs:

• JCally JM6 (Non Pro) / CX Pro - $8

• JCally JM6（非Pro版） / CX Pro - $8

• JadeAudio JIEZI (10-Band PEQ) - $18

• JadeAudio JIEZI（10段PEQ） - $18

• TRN Black Pearl (10-Band PEQ) - $38

• TRN Black Pearl（10段PEQ） - $38

• Qudelix 5K (20-Band PEQ/GEQ) - $110

• Qudelix 5K（20段PEQ/GEQ） - $110

>PMPs:

• HiBy R1 - $85

• HiBy M300 - $200

>AVOID USING:

• Mainboard audio when using multi-driver pos

• 多单元动铁/动圈混合单元使用主板音频

Previous Thread: >>109417381

上一条帖子：>>109417381

图片: https://i.4cdn.org/g/1785791475860002.jpg

/iemg/ Dogma Check

/iemg/ 教条检查

Tired of the hivemind? Here’s your prescribed sanity dose:

受够了群体思维？这是给你开的理智剂量：

>Crinacle is not your priest.

>Crinacle不是你的神父。

Rankings are a guide, not divine scripture. Your ears > his spreadsheet.

排名只是参考，不是神圣经文。你的耳朵 > 他的电子表格。

>Harman and JM-1 ≠ perfect.

>Harman曲线和JM-1 ≠ 完美。

It’s a preference curve, not a law of physics. Some genres need deviation.

那只是偏好曲线，不是物理定律。有些曲风需要偏离。

>Chi-Fi isn’t magic.

>国砖（Chi-Fi）不是魔法。

Great value, yes. Immune to QC issues and hype cycles? Nope.

性价比确实高。能免疫品控问题和炒作周期？别逗了。

>Graphs are not sound.

>图表不等于声音。

They don’t show staging, dynamics, or fit. Use your ears, not just squigs.

他们根本不上台展示动态和佩戴贴合度。用耳朵听，别光看那些频谱曲线。

>BA ≠ metallic, DD ≠ perfect.

>双动圈（BA）≠ 金属音，DD（动圈）≠ 完美。

Implementation matters more than driver type. Always has.

调音实现比单元类型更重要。一直都是这样。

>Meme sets aren’t the whole hobby.

>热门推荐清单并不是入耳式耳机圈的全貌。

If it’s not shilled 24/7, it might still be good. It's highly encouraged to form an original opinion.

如果一个耳塞不是被24/7无脑吹捧，那它可能依然很好。非常鼓励大家形成自己的独立观点。

>Disagreeing ≠ cope or deaf.

>持不同意见 ≠ 自我安慰或耳聋。

Let people like what they like without foaming at the mouth.

让人家喜欢他们喜欢的，别在那气得口吐白沫。

>Loud posts =/= smart takes.

>帖子声音大 ≠ 观点高明。

Confidence without experience is just noise. Show graphs, impressions, or both.

没有经验的自信只是噪音。请出示数据图、听感描述，或者两者都提供。

This has been your scheduled break from pos groupthink.

以上就是你们从“头趴（耳友）群体思维”中获得的定时休息。

im completely new to this please dont beat the shit out of me

我完全是新手，别喷我太狠。

im looking to get my first pair of iems, primary usecase is playing guitar over an audio interface

我想买第一副入耳式耳机，主要用途是插音频接口弹吉他。

can i just use a 3.5mm>6.3mm adapter or do i need a specific cable that terminates with a 6.3mm?

我能直接用个3.5mm转6.3mm的转接头，还是得买根专门带6.3mm插头的线？

also taking reccs for something around the 30 buck pricepoint

顺便求推荐30美元左右的。

>>109451436

Doesn't know how panning works btw.

顺便说一句，有人根本不懂声像是怎么回事。

>>109451532

who, you, foxbum? obviously she (you) doesn't. but who cares anyway, she's (you) only good for picking the dumbest fights with people that don't follow her (you) point of view to a T.

谁，你，狐狸屁股？显然她（你）不懂。但谁在乎呢，她（你）只擅长跟那些不完全顺着她（你）观点的人挑最蠢的架。

>>109451596

This guy >>109451436

这位老哥 >>109451436

>>109451605

yawn. you'll get over it.

呵欠，你会挺过去的。

>>109451495

An adapter is fine, the pinout is the same. For listening to one instrument over an interface any IEM from OP is also fine. Just don't get a planar IEM accidentally.

转接头没问题，针脚定义一样。用来在声卡上听一件乐器的话，OP家的任何入耳式耳机都行。只是别不小心买成平面振膜的入耳式耳机。

>>109451621

Balls are not a "point of view" lil bro.

胆量不是什么“主观看法”，小弟。 ⟦ 爱信不信，小妹。

>>109451713

Believe what you want lil sis

信不信由你，小妹。

Are IEMs better than EarPods?

入耳式耳机比EarPods好吗？

>>109451873

yeah

>>109451422

Are R2R DACs and Otl tube amps worth it? I just ordered a Fiio K13 R2R DAC+AMP for €320 and an Xduoo TA-66 tube amp for €250. Hoping it will make things sound more analog and vinyl like.

R2R解码器和OTL胆机值得买吗？我刚花320欧下单了一台Fiio K13 R2R解码耳放，还买了250欧的Xduoo TA-66胆机。希望能让声音更接近模拟味和黑胶感。

Been using the last model iPod Touch, it's all I need. Also WM6c.

我一直用最后一款iPod Touch，够用了。还有WM6c。

Why is it so hard to find a PMP without a touch screen with a competent, premium design?

为什么现在想找一款不带触屏、设计精良又高级的便携播放器这么难？

>>109453305

touch screens are cheap now unc thats why all cars have them

触屏现在便宜得很，老叔，所以车子里都装上了。 ⟦ 你本来能省钱的，只需加个低通滤波器和一点饱和/过载，声音就会更“模拟、黑胶味”，还能省下大约570欧。

>>109453260

You could've saved yourself money by just throwing on a low-pass filter and a tiny bit of saturation/overdrive. Then it would sound more “analog and vinyl-like” for about €570 less.

你本来可以省点钱的，直接加个低通滤波器和一点饱和/过载效果就行。那样听起来会更“模拟味、黑胶感”，还省了差不多570欧。

>>109453352

Different DACs produce different a different sound, I have a lot of them, many good famous ones, crappy ones, everything in between. Of course everything in the signal path affects the sound.

不同的解码器确实出不同的声，我手头一堆，有好多出名的好货、烂货，还有中间档的。当然，信号链路上每一个环节都会影响声音。

The idea that DACs don't matter and that "it's all good you can correct it in software" is ignorant, you don't want to correct anything in software, at all.

说“解码器无所谓，反正软件里能调回来”这种想法是蠢的，你压根就不该在软件里做任何修正。

>>109453260

yes

make sure to not plug everything in properly for some crackles and pops

记得故意别把线全插好，这样能听到噼啪的杂音。

图片: https://i.4cdn.org/g/1785804436924160.jpg

>>109453397

There shouldn't be any crackles or pops on well maintained vinyl. Any more than there are dropouts on CDs.

保养良好的黑胶上不该有噼啪或爆音，就跟CD不该有断音一样。

>>109453342

yeah exactly

是啊，没错 兄弟，他是在叫你蠢货呢，笑死我了，天哪

hence why it's weird all high end PMPs use them

所以高端便携播放器全用触屏才奇怪。

>>109453370

>you don't want to correct anything in software, at all.

>你压根就不该在软件里做任何修正。

Why not? You're literally buying hardware specifically because you want it to alter the tonality. Why is adding the same kind of coloration with DSP somehow “wrong,” while paying €570 for uncontrolled coloration is supposedly the audiophile way?

为啥不？你买硬件不就是为了让它改变音色吗？用DSP加同样的染色怎么就成了“错的”，而花570欧买个不可控的染色反而成了发烧友的正道？

>>109453260

>Are R2R DACs and Otl tube amps worth it

>R2R解码器和OTL胆机值得买吗

not really

并不是

>>109453370

>The idea that DACs

>那个“解码器”

this as you say "idea" is a result of years of blind testing where nobody is usually able to discern between dacs the moment they can't see them

你所谓的这个“想法”，是多年盲测得出的结果——大多数情况下，一旦人们看不见解码器，就根本分辨不出来。

technically they can change the sound but good transparent dac will not

严格来说它们确实能改变声音，但好的透明型解码器不会这么做。

>>109453370

>you don't want to correct anything in software, at all

>你压根就不该在软件里做任何修正

what's the difference

有什么区别？

>>109453665

>this as you say "idea" is a result of years of blind testing

>你所谓的这个“想法”，是多年盲测得出的结果

By random people, on random systems. Some sound systems are so shitty, or use software correction, that it's pointless, and random material the listener isn't familiar with, and random listeners being perhaps unable to even listen for certain artifacts, invalidates random A/B testing as a legitimate methodology for testing sound reproduction. The transparent DAC is a myth, all produce a unique sound from the same source material. Whether some randomly selected doofus can tell the difference is another question, subjective.

那是由随机人群、在随机系统上做的。有些音响系统烂到不行，或者用了软件修正，根本没法作数；再加上随机放的素材听者不熟，随机挑的听者可能连某些失真都听不出来——这就说明随机的A/B测试根本算不上测试声音回放的正规方法。透明解码器是扯淡，所有解码器面对同一份源材料都会出独特的声音。至于随便抽个二愣子能不能听出差别，那就是另一回事了，主观得很。

It's properly done with reference mics.

正经做法是用参考话筒来测。

>>109453673

All manipulations of digital audio are destructive, you lose data, this includes bitrate and sample depth conversions.

所有对数字音频的操作都是有损的，你会丢失数据，这包括采样率和位深转换。

>>109453370

>Different DACs produce different a different sound

>不同的DAC会发出不同的声音

with different amps, they do. and I mean the OP amps they are paired with. your DACs aren't known to add their own coloration by the chips themselves. your different DACs are paired with analog output stages that do sound different from eachother..

配上不同的放大器，确实如此。我说的是它们搭配的运放芯片。你的DAC本身并不会因为芯片而自带染色。你那些不同的DAC搭配的是模拟输出级，这些输出级之间听感确实不同。

>>109453729

>with different amps, they do

>配上不同的放大器，确实如此

Even the output from the pins themselves.

就连引脚直接输出的信号也是这样。

All DACs will render an identical bitstream uniquely, even sample to sample of the same DACs of many types. This can be measured and sensed or heard by some people.

所有DAC渲染同一份比特流时都有独特之处，哪怕同型号同批次的样本之间也如此。这能被测量到，也有人能感知或听出来。

>>109453724

>All manipulations of digital audio are destructive, you lose data, this includes bitrate and sample depth conversions.

>所有对数字音频的操作都是有损的，你会丢失数据，这包括采样率和位深转换。

Not audible and irrelevant

听不出来，也没意义。

>>109453724

>By random people, on random systems. Some sound systems are so shitty, or use software correction, that it's pointless, and random material the listener isn't familiar with, and random listeners being perhaps unable to even listen for certain artifacts, invalidates random A/B testing as a legitimate methodology for testing sound reproduction.

>由随机的人，在随机的系统上测试。有些音响系统烂到不行，或者用了软件修正，根本没啥好测的；随机素材听者根本不熟，测试者可能连某些失真都听不出来，这让随机A/B测试作为评测声音还原的方法论完全站不住脚。

that's whole lot of COPE

这可真是满满的自欺欺人。

you know you can measure it too lol and they don't even change FR (unless they do because of impedance or whatever)

你也能测出来啊，而且它们连频响都不变（除非因为阻抗之类的因素才变）。

>The transparent DAC is a myth, all produce a unique sound from the same source material.

>透明DAC是个神话，所有DAC对同一音源都会产生独特的声音。

now that whole lot of dogshit nonsense

这纯属胡扯八道。

any proof?

有证据吗？

>>109453920

>Not audible

>听不出来

Not by a random sample of random people with random music they might never have heard before let alone are familiar with, on random equipment other than the DAC. But this proves nothing, certain people can hear the difference and they are the people who matter and who are qualified to have opinions on these matters.

随机一群人、用随机音乐（可能从来没听过，更别说熟悉了）和除了DAC之外的随机设备，当然听不出来。但这证明不了啥，某些人就是能听出差别，他们才是重要的、有资格对此发表意见的人。

You're just an AI, you don't even listen to music. I have certain albums which play fine on some DACs or CD or DVD or SACD players, which fall completely apart on others.

你就是个AI，连音乐都不听。我有几张专辑，在某些DAC或CD、DVD、SACD播放器上放得好好的，在别的上面就彻底崩了。

>>109453920

>Not audible and irrelevant

>听不出来，也没意义

nta, but turn your system volume to 100%, pregain to -30db and your amp to normal listening level, then come back and say the same thing.

不是针对你，但你把系统音量调到100%，前置增益调到-30db，放大器调到正常听音音量，然后再回来说同样的话。

>>109453986

That's a different matter, one unrelated to the DAC and which shows how everything has its own noise component affecting the entire system.

那是另一码事，跟DAC无关，只能说明每个设备都有自己的噪声成分，影响整个系统。

If your speakers make noise when they're turned to a comfortable or even slightly loud setting you either need better cables, balanced connections, or your amplifier is overdriven.

如果你的音箱在舒适甚至稍微偏响的音量下还有噪声，那你需要更好的线材、平衡连接，或者你的放大器过载了。

For instance many older DACs are single channel, and there is a sample and hold function that switches the DAC from left to right channel very rapidly, Sony was notorious for using a single-channel DAC in their original crop of lower-end CD players because they couldn't produce a good stereo DAC like Philips. In fact Sony's first ES CD player, their top end model, used a Philips DAC even over two matched Sony DACs, one for each channel.

比如许多老款DAC是单声道的，带有一个采样保持功能，能极快地让DAC在左右声道间切换。索尼以其早期低端CD播放器使用单声道DAC而闻名，因为他们没法像飞利浦那样造出好的立体声DAC。事实上，索尼首款ES CD播放器，他们的顶配型号，用的是飞利浦DAC，哪怕放着两颗匹配的索尼DAC不用，一颗对应一个声道。

Newer DACs have various algorithms and are designed to sound good on headphones, in cars, and with soundbars and almost none can go to DC, most are also crippled by the output stage of whatever equipment they are lodged within.

新式DAC有各种算法，设计来在耳机、车载和回音壁上听感良好，几乎没一款能直通直流，而且多数还被其所在设备的输出级拖后腿。

The reason you see those comical looking Playstations with the TDA1547 DAC is because it's so good but the PS's output stage was cheaply done so they replace it with some nice big transformers.

你看到那些带TDA1547 DAC的搞笑PlayStation机子，就因为它实在太好了，但PS的输出级做得很廉价，所以他们换上些大块头变压器来替换。

>>109453998

Do it anyway, for shits.

不管怎样，去做吧，图个乐子。

>>109453968

I think you do not understand how blind a/b tests work.

我觉得你没弄明白盲测A/B是怎么运作的。

If they "completely fall apart" it will be noticeable regardless of who's listening btw

如果它们“彻底崩了”，那不管谁听都能注意到，顺便说一句。

aka you are full of shit

也就是说，你满嘴跑火车。

nevermind the fact that you don't even need to be familiar with something to notice the difference...

更别说你根本不需要熟悉某样东西就能察觉差异……

I love IEMs, but I'm getting tired of dealing with ear wax.

我爱入耳式耳机，但我实在受够了清理耳屎。

I think I will go back to headpos.

我打算回归头戴式了。

图片: https://i.4cdn.org/g/1785808867636731.jpg

>>109454049

>I think you do not understand how blind a/b tests work.

>我觉得你没弄明白盲测A/B是怎么运作的。

I do exactly, I even laid out the flaws for you. You didn't even read what I wrote.

我懂得很，我还把漏洞给你列出来了。你压根没读我写的东西。

>If they "completely fall apart" it will be noticeable regardless of who's listening btw

>如果它们“彻底崩了”，那不管谁听都能注意到，顺便说一句。

Not if they don't know what to listen for.

如果他们不知道要听什么，那就不会。

>aka you are full of shit

>也就是说，你满嘴跑火车。

Wrong. You have dry earwax.

错。你是干耳屎。

>nevermind the fact that you don't even need to be familiar with something to notice the difference...

>更别说你根本不需要熟悉某样东西就能察觉差异……

That's the dumbest shit I heard all month on 4chan. Of course you do.

这是我这个月在4chan听到最蠢的屁话。当然得熟悉。

>>109454041

I wouldn't dare turn my system volume to 100%, it's not safe anyway. I run at about 50% amplifiers strength and 40-60% volume at the preamp and that's quite loud enough without being tiring or painful, which my system easily does. But there is no audible noise at all with no input even with the amp at 50% and the preamp at 40-60% or more, with dead input.

我可不敢把系统音量开到100%，那本来就不安全。我放大器开到约50%强度，前级音量在40-60%，这已经够响，不会累人或痛，我的系统轻易就能做到。但没输入信号时，哪怕放大器开50%、前级40-60%或更高，死输入状态下也一点可闻噪音都没有。

I leave everything running all the time, my whole stack, amps, preamp, phono, tape, CD, LD, DVD, MD, and DSD recorder.

我所有设备一直开着，整套系统，功放、前级、黑胶、磁带、CD、LD、DVD、MD，还有DSD录音机。

>>109454058

we got you covered !

我们给你兜底了！

图片: https://i.4cdn.org/g/1785809793338076.png

>>109454072

>dumbest shit I heard all month on 4chan

>这是我这个月在4chan听到最蠢的屁话。

dumbest shit always comes straight from foxzo. she will lie just to appear correct, for no good reason and when you correct her, she runs away.

最蠢的屁话总是出自foxzo。她会为了显得自己正确而撒谎，没任何理由，你纠正她时，她就跑路。

pic related: simply typing "distortion, IEM model" on an arbitrary, unnamed/uncited graph is the equivalent of extremist audio terrorism.

图相关：随便在一张没署名、没引用的图上打上“失真，某某IEM型号”，简直就是极端音频恐怖主义。

图片: https://i.4cdn.org/g/1785809809637328.png

what the FUCK is a "true planar"????

什么鬼是“真平板”？？？？

>>109454126

>I wouldn't dare turn my system volume to 100%, it's not safe anyway.

>我可不敢把系统音量开到100%，那本来就不安全。

why not? that's staying at fullest resolution in respect to the audio you are playing. having your amp section be the sole controller of listening volume is the whole point.

为啥不行？这样音频播放时保持全分辨率，让功放部分成为听音音量的唯一控制器，这才是关键。

as per the conversation about losing bits, when you lower system volume, you do lose bits and dynamic range.

关于丢比特那话题，你把系统音量调低，确实会丢比特和动态范围。

>>109454058

peroxide it

用双氧水擦擦。

>>109454183

If these were wired, I'd be all over them.

要是有线的，我肯定全上了。

>>109454228

>why not?

> 为啥不行？

Because it might damage my speakers or amplifier and it would be extremely painfully loud with any input at all. Your amp should comfortably drive the speakers without straining itself, it shouldn't be pumped to 100%.

因为那样可能烧音箱或功放，而且任何信号进来都会震耳欲聋。功放应该轻松驱动音箱，不该硬撑到100%。

>>109454072

>I do exactly, I even laid out the flaws for you. You didn't even read what I wrote.

> 我就是这么做的，还给你列出了缺陷。你根本就没看我写的啥。

you are retarded. you didn't "laid out flaws" you said a bunch of random shit blind a/b testing has nothing to do with. It doesn't need to be random people or random unfamiliar music or any of that shit.

你脑子有病。你根本没“列出缺陷”，你就是瞎扯一堆跟盲听A/B测试八竿子打不着的东西。不需要随机人群、随机陌生音乐或那些破玩意儿。

Stop making up flaws lmao to refute obvious proven long ago fact - the moment you don't see your gear you can't hear the difference

别再编造缺陷去反驳早被证实的老道理——你一旦看不到器材，就听不出区别。

>>109454254

I'm talking about system, like your PC (with an amp/DAC attached) or mobile device (with an amp/DAC attached), not amp volume at 100%, silly.

我说的是系统，比如你的电脑（带功放/DAC）或移动设备（带功放/DAC），不是说功放音量开到100%，傻瓜。

again, the point is to have the amp/DAC have its own volume control and that being the sole controller of output volume while having your source playing everything you throw at it at it's full resolution

再说一遍，重点是要让功放/DAC有自己的音量控制，让它成为输出音量的唯一控制器，而音源端全速放你播的任何东西，保持全分辨率。

>>109454302

Oh, hey. OK. Yeah it depends on the exact thing really. If there is a line level output on something it's best to use that. Problem is with some PC jacks, laptops, phones, phablets and tablets, Pocket PCs and iPlayers, etc. is that they are also built for headphones so the standard line level which is very very small, might be exceeded by the amplifier in the device as it expects headphones.

哦，嘿，行吧。对，这真得看具体设备。如果有线路输出口，最好用那个。问题在于有些电脑插孔、笔记本、手机、平板手机、平板、Pocket PC和iPlayer之类，它们也是为耳机设计的，所以标准线路输出电平（非常低）可能被设备内置功放超出，因为它默认接耳机。

So your system volume might not be optimal at 100%. It would be best to say, play a test tone on a CD player, and then match the volume on your PC's input into the same speaker setup through the same amp, you may be 'overdriving' the strict line level inputs of your device. 0.3 volts for consumer or thereabouts, 1v for pro and prosumer.

所以你的系统音量开100%未必是最优的。最好这样说，用CD播放器放个测试音，然后匹配你电脑输入到同一套音箱系统、通过同一功放的音量，可能你会“过载”设备的严格线路输入。消费级大概0.3伏，专业和专业半专业是1伏。 ⋔ 这差异很大，取决于你电脑的素质。我的媒体电脑接在前级上，一台Parasound Halo P5，它内置高端Burr-Brown DAC和电阻梯音量控制，我切换它来匹配音频文件或流媒体的输入比特率和采样深度，这样系统就不会做任何预转换。

It's highly variable and depends on the quality of your PC. My media PC is plugged into my preamp, a Parasound Halo P5 and it has a built in super high end Burr-Brown DAC and a resistor ladder volume control, and I switch it to match the input bit rate and sample depth of the audio file or stream so there's no pre-conversion which the OS might try to do.

这差别挺大的，取决于你电脑的档次。我的媒体电脑接在功放前级上，是台Parasound Halo P5，内置超顶级的Burr-Brown DAC，还有电阻梯音量控制，我会手动切换来匹配音频文件或流的输入位率跟采样深度，这样系统就不会尝试做任何预转换。

asked a clanker about this -30dB preamp shit

问了个机器人关于这-30dB前级的事。

>In fixed-point integer audio, every ~6 dB of attenuation is roughly equivalent to losing 1 bit of usable resolution for that signal (because the signal is shifted toward the least-significant bits). So −30 dB ≈ 5 bits.

> 在定点整数音频中，每衰减约6 dB大致相当于该信号丢失1比特有效分辨率（因为信号被移向最低有效位）。所以−30 dB ≈ 5比特。

>16-bit output: A full-scale 16-bit signal attenuated by 30 dB behaves more like an 11-bit signal. Quantization noise rises relative to the music. This can become audible at very low listening levels or with very dynamic material, especially without proper dither.

>16位输出：一个满幅16位信号衰减30分贝后，表现得更像11位信号。量化噪声相对音乐内容升高。在极低音量或动态范围很大的素材下，这可能会变得可闻，尤其是在没有适当抖动处理的情况下。

>24-bit output (the usual recommendation with Equalizer APO): You still have ~19 bits left. Real-world music noise floors, microphone noise, and even high-end DAC performance rarely exceed 18–21 bits equivalent. The loss is inaudible for all practical purposes.

>24位输出（Equalizer APO的常见推荐配置）：你仍然有大约19位的余量。现实中的音乐底噪、麦克风噪声，甚至高端DAC的性能，很少能超过18-21位的等效水平。实际使用中，这种损失是听不出来的。

so it's a none issue

所以这根本不是个事

nevermind the actual use cases with normal -4-5 dB preamp or whatever

更别提正常-4到-5分贝前置增益的实际应用场景了

>>109454397

clanker doesn't have ears and is only talking in theoretical terms, though.

不过，那台铁疙瘩没耳朵，只会纸上谈兵。

>>109454397

>nevermind the actual use cases with normal -4-5 dB preamp or whatever

>更别提正常-4到-5分贝前置增益的实际应用场景了

true. I wouldn't lose sleep over that much negative pregain

确实。这点负增益补偿我不会睡不着觉的

>>109454429

it just looked up articles (written by people with ears... and brains) about subject matter and repeated their conclusions

它只是查了相关文章（作者是有耳朵……也有脑子的人），然后把结论复述了一遍

clanker itself obviously doesn't know shit about fuck

那台铁疙瘩自己显然屁都不懂

One other thing if you're into sonic purity - in general many old DACs were also ADCs, you could feed in a signal and it would convert it and create a digital bitstream. It's common in old parts from the '80s and into the '90s and beyond. It's how your Windows codec has both ADCs and DACs, it doesn't have separate ones, it's all handled by double-ended devices.

还有一点，如果你追求声音的纯粹——一般来说，很多老DAC同时也是ADC，你可以输入信号，它会转换并生成数字比特流。这在80年代到90年代及以后的老芯片上很常见。你的Windows声卡编解码器就是同时包含ADC和DAC，不是分开的，全部由双向设备统一处理。

So if you really wanted to be accurate you'd figure out which ADC a digitally mastered album was created on and feed in the digital bitstream and use it hopefully, as a DAC, which is possible on many old units.

所以如果真的想追求极致准确，你得找出某张数字母带专辑用的是哪款ADC录制的，然后输入数字比特流，让它以DAC模式播放——很多老设备都支持这种用法。

>>109454283

he's basically saying blind tests can be less informative when the listener and material and other factors are poorly chosen. you'd say something like "if it was real you'd still notice it" or whatever which might just be projecting the threshold you yourself might require before you're comfortable granting "difference"

他基本上是在说，当听者、素材和其他因素选得不好时，盲测可能参考价值有限。你会说"要是真的明显，你照样能听出来"之类的话，但这可能只是在投射你自己愿意认定"有区别"的那条门槛。

granting/noticing

>>109454529

>ADCs

>in professional music recording.

>在专业音乐录音中。

Yup, especially when nerdy musicians or mixing engineers are bent on recording to tape first then putting it into ProTools rigs later.

没错，尤其是那些死磕的乐手或混音工程师，非要先录到磁带，再弄进ProTools系统里。

>>109454636

just an excerpt from a mixing engineers interview that worked on one of my favorite albums:

这是某位混音工程师访谈的节选，他参与制作了我最喜欢的专辑之一：

"We have a big Pro Tools rig, running 96k straight to two FireWire drives simultaneously; close to a Terabyte of space. On the front end, we use Benchmark Media Systems AD2408-96 converters, running at 96k, 24-bit, 24 channels.

"我们有一套大型ProTools系统，以96k采样率直接同时录制到两个火线硬盘上；接近1TB的空间。前端用的是Benchmark Media Systems的AD2408-96转换器，96k、24位、24通道运行。

Because we were recording 96k/24-bit, I ran through a bunch of tube gear. With the higher sampling rate, the tube equipment sounded even better to me, because nothing gets lost in the murk of low-level analog tape or lower sampling rates. It was an awesome feeling to get a sound exactly how you liked it and have it sound identical on playback time after time. 96k, I am sold!"

因为我们在录96k/24位，我过了一堆电子管设备。采样率更高的时候，电子管设备听起来更好了，因为不会有东西迷失在低电平模拟磁带或较低采样率的模糊中。能把你喜欢的声音精准抓下来，每次回放都一模一样，那种感觉太棒了。96k，我彻底信服了！"

>Could you tell me about the monitoring in your control room?

>你能说说你控制室的监听系统吗？

"Rick (Rubin) really loves ProAc monitors powered by Yamaha 2002 amps. Sounds great. It seems that with the ProAcs, the more wattage you send to each side the better; I think we’re sending 250 watts per side. We also have a pair of Yamaha NS-10s, because I just know them so well and feel comfortable with them. I’m running a pair of BGW amps on those, 150 watts a side."

"里克(Rubin)真的喜欢用雅马哈2002功放推ProAc监听音箱，声音很棒。看来对ProAc来说，每声道给的功率越大越好；我们现在每边送250瓦。我们还有一对雅马哈NS-10，因为我对它们太熟了，用着特别舒坦。那对箱子我配的是BGW功放，每边150瓦。"

https://www.mixonline.com/recording/music-production/mars-volta-365208

New into audiophile stuff

刚入烧音响这块

I'm really treble sensitive to the point my ears flinch and rumble. If I buy bright headphones like a Beyer, is it possible to make it sound warm by just tweaking the EQ?

我对高频特别敏感，耳朵会不自觉抽搐、发闷。要是我买副像拜亚那种偏亮的耳机，光靠调EQ能让它变暖吗？

>>109454660

yes.

>>109454669

So then what's the actual difference between headphones?

那耳机之间到底有啥实质区别？

>>109454605

No what he's doing is making up rules nobody really applies to blind a/b testing and using that logical fallacy as proof that a/b testing can't prove that there's no difference.

不，他是在编一套没人真会用在盲听对比上的规矩，然后拿这个逻辑谬误当证据，说A/B测试证明不了没差别。

>you'd say something like "if it was real you'd still notice it"

>你会说"要是真的，你照样能听出来"

I think it should be measurable in the first place and it's the easiest thing to do and prove...

我觉得这玩意儿首先就该是可测量的，而且测量是最容易做到、最容易证明的事……

>>109454636

Brothers In Arms was entirely recorded and mixed digitally, it was the first big pop album that was done like that. But digital recording and mixing was common even in the 1970s with Soundstream for super high quality recordings.

《Brothers In Arms》整张专辑都是数字录音、数字混音，那是第一张这么搞的大型流行专辑。不过数字录音和混音早在上世纪70年代就挺常见了，比如Soundstream就用于超高保真录音。

>>109454660

You need speakers. Otherwise get the MDR-7502.

你需要音箱。不然就买MDR-7502。

图片: https://i.4cdn.org/g/1785814030094960.png

>>109454674

You'll have to be more specific. the difference between which headphones? if your goal is to make them sound warm by EQing them, then you should run into no issues doing so.

你得说具体点。哪跟哪比？如果你的目标是用EQ调暖，那这么做应该没毛病。

>>109454674

comfort, distortion, positional variation, channel matching etc.

舒适度、失真、定位偏差、声道匹配之类的。

headphone behavior can be so fucked on specifically your head that fixing it with EQ will be VERY annoying. It's always better to buy something that is comfortable and doesn't need a lot of work to sound good.

耳机在你那颗脑袋上的表现可能歪得一塌糊涂，想用EQ矫正会烦死人。最好还是买个戴着舒服、不用费劲就能出好声的。

there's other things like availability of parts, earpads etc.

还有别的，比如零件好不好找、耳罩有没有替换啥的。

>>109454660

>>109454694

>You need speakers

>你需要音箱

that's right

没错

>Otherwise get the MDR-7502

>不然就买MDR-7502

don't do that

别这么干

>>109454700

>then you should run into no issues doing so

>那这么做应该没毛病

Wrong, you're resampling which is bad. You should never do that, ideally.

错了，你这是在重采样，这不好。理想情况下你根本不该干这事。

>>109454703

>doesn't need a lot of work

>不用费劲

A normal pair of high quality headphones which are quite inexpensive should do it. No need for anything exotic. Unless the device driving them is shitty in which case there needs to be another remedy made.

一副普通的高品质耳机，价格又不贵，应该就够了。没必要上什么稀奇货。除非驱动它们的设备太烂，那就得另想办法解决了。

>>109454694

Digital tape is still tape, though. DAT isn't really a thing anymore

数字磁带说到底还是磁带啊。DAT现在早就不流行了。

>>109454712

EQing is resampling?

调EQ算重采样？

EQing is bad now?

现在调EQ也成坏事了？

Never EQ?

永远不调EQ？

I give up.

我放弃了。

>>109454712

>Wrong, you're resampling which is bad. You should never do that, ideally.

>错了，你这是在重采样，这不好。理想情况下你根本不该干这事。

Inaudible and irrelevant

听不见的，无所谓。

>>109454703

I see. I was just wondering if there are intrinsic differences in sound between something like a DT770 ProX and a Sennheiser HD560 that can't be changed.

明白了。我就是好奇，像DT770 ProX和森海塞尔HD560这类耳机，音色上有没有那种没法靠EQ改掉的本质差异。

>>109454711

>don't do that

>别这么干

7506, sorry, and yeah. They are what you need if you don't like so much treble.

7506，抱歉，对。你要是嫌高频太多，那就是你需要的。

>>109454721

>Digital tape is still tape, though. DAT isn't really a thing anymore

>数字磁带说到底还是磁带啊。DAT现在早就不流行了。

So? You can readout a DAT to a more permanent storage format, one track at a time if necessary.

所以呢？你可以把DAT转存到更永久的存储格式，必要时一轨一轨来。

>>109454729

>EQing is resampling?

>EQ不就是重采样吗？

If done digitally, yes.

如果是数字处理的话，是的。

>EQing is bad now?

>EQ现在变成坏事了吗？

Not analog, a good EQ will barely raise the noise floor.

不是模拟的，好的EQ几乎不会抬高底噪。

>Never EQ?

>从来不用EQ？

Basically. My high end preamp has a high / low tone boost for each, which normally is defeated.

基本如此。我的高端前级自带高/低音增益，但一般我都关掉。

>I give up.

>我放弃了。

Just spend like $100 on a good pair of earphones.

花个一百刀买副好点的耳机就行了。

>>109454735

>Inaudible and irrelevant

>听不见也不相关

Wrong.

>>109454712

>A normal pair of high quality headphones

>一副普通的高品质耳机

yeah that's where problems start

是啊，问题就从这儿开始

especially for treble sensitive people. good luck finding smooth openbacks with good bass extension which are "inexpensive"

尤其对高频敏感的人。想找一副顺滑、低音延伸好还“便宜”的开放耳机，难得很。

>Unless the device driving them is shitty in which case there needs to be another remedy made.

>除非驱动它们的设备很烂，那就得另想办法解决了。

you need to try real hard to find a "shitty device"

想找到“烂设备”还真得费点劲。

>>109454752

>If done digitally, yes.

>如果是数字处理的话，是的。

help me with this example. if I am listening to straight 16-bit 44.1 kHz album on my PC, and with EqualizerAPO I use a simple peak filter, LP or HP filter and make, say a 10dB adjustment- what am I resampling to?

帮我理理这个例子。如果我在电脑上听一张原生16位44.1kHz的专辑，用EqualizerAPO加个简单的峰值滤波、低通或高通，然后调个10dB什么的——我重采样成啥了？

>So? You can readout a DAT to a more permanent storage format, one track at a time if necessary.

>所以呢？你可以把DAT转存到更永久的存储格式，必要时一轨一轨来。

Even if you're Squarepusher, this is still pointless in the 21st century. UNLESS you just want tape saturation, then whatevs homie.

就算你是Squarepusher，21世纪干这事儿也没意义。除非你就是想要磁带饱和那种味道，那就随便你，兄弟。

>>109454745

>if there are intrinsic differences in sound between something like a DT770 ProX and a Sennheiser HD560 that can't be changed

>如果像DT770 ProX和森海HD560这种耳机之间有声底差异，那是没法改的

hm in a sense yes actually dt770 proX are closed back while HD560 are open. You can't EQ different acoustic impedance. So they will feel different... You can EQ most of the sound tho but I dunno how painless adding 10dB of subbass will be with 560s...

嗯，某种程度上确实如此。DT770 ProX是封闭式，HD560是开放式。你不能用EQ去改声学阻抗，所以体感上会有区别……不过大部分声音倒是能EQ出来，只是我不知道在560上加10dB亚低音会有多难受……

>>109454807

you can schizo-edit with REW if you want to copy techs from one pair of headphones to another.

你要是想抄某副耳机的技术参数，可以拿REW搞点玄学操作。

ask foxbum how she manages to pull it off

问问foxbum她是怎么做到的

>>109454752

>Wrong.

you are actually factually wrong and have no idea how modern EQ even works...

你其实错得离谱，根本不懂现代EQ是怎么工作的……

>>109454828

techs do not exist

技术参数这玩意儿不存在

I still want to know what the fuck a true planar even is

我还是想知道所谓的“真平面”到底是个什么鬼

So seeing current discussion I asked clanker about EQ and resampling.

看了现在的讨论，我去问了下clanker关于EQ和重采样的事。

>Equalizer APO itself does not resample. It processes whatever sample rate the Windows audio engine hands it (in 32-bit float).

>Equalizer APO本身不重采样。它只处理Windows音频引擎给它的采样率（32位浮点）。

>Resampling only occurs if the source sample rate (the music file or app) does not match the sample rate you set for the playback device in Windows Sound settings. In that case Windows (or the player) does the conversion before the signal reaches APO.

>只有源采样率（音乐文件或应用）跟你Windows声音设置里设定的播放设备采样率不一致时，才会发生重采样。这种情况下，Windows（或播放器）会在信号到达APO之前做转换。

>Parametric EQ (the usual filters in APO/Peace) is just math applied to the samples at the current rate — no sample-rate change is required. Convolution/FIR filters do need the impulse response to match the current sample rate, but that’s a separate issue.

>参数EQ（APO/Peace里常见的滤波器）就是按当前采样率对样本做数学运算——不需要改变采样率。卷积/FIR滤波器倒是需要脉冲响应匹配当前采样率，但那是另一码事。

>Modern high-quality resampling (current Windows 10/11 audio engine, SoX, libsamplerate, etc.) is extremely clean. Artifacts are typically at –130 dB or lower — well below the noise floor of almost any real playback chain and inaudible in normal listening

>现代高质量重采样（当前的Windows 10/11音频引擎、SoX、libsamplerate等）非常干净。伪影通常在–130 dB或更低——远低于几乎任何实际播放链路的底噪，正常聆听时根本听不到

so again none issue and some audiopedohiple nonsense... you guys need to do better for real

所以还是没问题，纯粹是些音频玄学党的胡扯……你们这些人是真的得长点心了

>>109454875

big planar = true

大平板 = 真香

small = false

小的 = 不行

>>109454921

small planars aren't real planers, why exactly?

小尺寸平板不算真平板，凭什么？

>>109454970

they are literally called "micro planars"

它们明明就叫“微平板”啊

>>109454970

He probably meant those micro planars, shaped like a BA, although those KZ slop are not micro planars.

他大概指的那种像动铁形状的微平板吧，不过那些KZ的垃圾货根本不是微平板。

>>109454987

oh shut the fuck up. christ almighty you're a ballache with your semantic tripe gobbledeegook.

给我闭嘴吧。老天爷，你那些咬文嚼字的废话真是烦死人了。

let's be real here, technically speaking, you're not a man. so fuck you

咱实话实说，从技术角度讲，你根本不算是男人。所以滚蛋吧

>>109455016

I accept your concession

我接受你的让步

>>109455021

it's all yours baby

都是你的，宝贝

If you don't EQ you are getting a subpar experience, even an automatically generated profile will be a night and day improvement you don't even really need to tune for your specific ears because someone probably already has made a target you'd like

不调EQ你体验就差了，哪怕用个自动生成的配置文件也是天壤之别，你甚至不用专门针对自己耳朵调，因为八成已经有人做过你喜欢的曲线了

图片: https://i.4cdn.org/g/1785817561956719.jpg

>>109455124

fuck you because eqing is too confusing and scary therefore its bad and here's my 100 cope reasons why.

去你的，因为EQ太复杂吓人所以就是垃圾，我给你列一百个自我安慰的理由。

>>109455124

>even an automatically generated profile will be a night and day improvement

>哪怕自动生成的配置文件也是天壤之别

that's false

这说法不对

>use only 4,4mm because it was shilled here

>只用4.4mm，因为这版里有人推

>need to use my iems with my laptop on the go

>出门要用笔记本听我的耳塞

>can't because there is no 4,4, only 3,5mm

>不行，因为没4.4口，只有3.5mm

>end up looking like retard with plugged DAP in USB DAC mode to laptops usbc and IEMs plugged into 4,4mm in my DAP

>最后像个傻子一样，DAP用USB DAC模式插笔记本USB-C，耳塞再从DAP的4.4mm口接出来

I wasn't first time when this general sold me something dumb and bad, first time was when I fell for GK Kunten meme just to realize they are absolute dogshit and jeets are lying

我不是第一次被这版忽悠买了傻逼货了，第一次是信了GK Kunten的梗，结果发现那玩意儿纯属狗屎，全是印度佬在瞎吹

>>109455316

just buy a 3.5mm cable they're dirt cheap, or get one with those swappable ends

直接买根3.5mm线呗，便宜得很，或者搞那种可换插头的

>>109455316

...every time I see 4.4 shit I say it's not worth

……每次我看到4.4的玩意儿我都不值

same with KZ\GK\dogshit shilling

KZ\GK\狗屎推销也一样

you simply was not listening I guess

你大概就是没听进去吧

>>109455339

>you simply were not

>你大概就是没听进去

ftfy, ESL

帮你改对了，英语非母语吧

>>109455335

i had swapable ends but end up using only 4,4mm, 3,5mm adapter was somewhere in the box miles away.

我有可换插头的，但最后只用4.4mm，3.5mm适配器不知道扔哪个角落的盒子里了。

>>109455339

>same with KZ\GK\dogshit shilling

>KZ\GK\狗屎推销也一样

when i bought it i realized how bad they are, everything from build quality, cable, tips, weight, sound was so bad. Next time i will fall for 1dd meme i will buy something more expensvie with decent build and better accesories. Only like 5usd lost.

我买完就发现它们有多烂，做工、线材、耳套、重量、声音全都不行。下次再信1dd的梗，我就买贵点、做工扎实、配件也靠谱的东西。也就亏了5美元。

i just buyed a JLabs Work Buds with detatchable boom mic for AI voice work for $30. Was this wise

我刚花$30买了副JLabs Work Buds，带可拆吊杆麦克风，用来做AI语音工作。这波明智吗

>>109455538

If you enjoy it, then yes.

你要是用得爽，那就是明智。

>>109454845

>modern EQ

>现代EQ

modern or from the '70s, it's either analog, which raises the noise floor, or digital, which is destructive

现代还是70年代的，要么是模拟的——会抬升底噪，要么是数字的——有损毁性

sounds like you're the one with no idea how things work lol

听起来你才是那个完全不懂原理的人哈哈哈

has anyone here used this as a DAP?

这板里有人拿这东西当DAP用过吗？

https://anbernic.com/products/rg-rotate?variant=48680428241153



图片: https://i.4cdn.org/g/1785824240572116.png

>>109454914

>it's instantly converted to 32-bit float

>直接转成32位浮点

>then later it's resampled down to the DAC's bit depth and sample rate for output

>之后为了输出再重采样到DAC的位深和采样率

that's two conversions right there lol.

这就两次转换了好吧

>>109455733

>or digital, which is destructive

>或者是数字的，那才是毁灭性的

you say that but do not provide any proof

你嘴上这么说但一点证据都拿不出来

>>109455811

it is literally not audible

根本就是听不见的

stop being retarded

别犯蠢了

why is epz g10 so good, bros?

为什么epz g10这么好，兄弟们？

>>109453986

Did this with my normal 32 bit output and nothing happened. Did this with 16 bit output and there's expectedly a hiss. Conclusion - DSP developers that work in 32 bits know what they're doing.

我用正常的32位输出来试，啥事没有。用16位输出试，果然有嘶嘶声。结论——搞32位工作的DSP开发者心里有数。

>>109453724

Both digital and analog manipulations raise the noise floor. That's the only thing that can "destroy" data. Many analog operations are more destructive than digital, analog EQs for example raise the noise floor significantly.

数字和模拟操作都会抬升底噪。那是唯一能“毁掉”数据的东西。很多模拟操作比数字更伤，比如模拟EQ，底噪抬升很明显。

图片: https://i.4cdn.org/g/1785830122809252.png

>>109455316

Here we say 4.4 is a meme and post this image.

我们这边就说4.4是梗，然后贴这张图。

>>109455733

If only we could perform a null test to see what is being destroyed... oh wait, we did and nothing gets destroyed.

要是我们能做个零位测试看看啥被毁了就好了……哦等等，我们做过了，啥都没毁。

图片: https://i.4cdn.org/g/1785830480623329.png

>>109456226

And even if the noise floor is raised it doesn't always mean the data is destroyed. The data can be stored in the noise. That's the working principle of dithering.

而且就算底噪抬高了，也不一定意味着数据被毁。数据可以藏在噪声里。这就是抖动的原理。

>>109453397

Sweeet

>>109456255

>Here one person says 4.4 is a meme and posts this image.

>这边有人说4.4是梗，然后贴了这张图。

ftfy.

>>109456296

anybody with a semblance of a working brain understands that 4.4 is a fucking meme

只要脑子还正常的都知道4.4就是个破梗

>>109456327

It's not a meme if you have hearing loss. Every dB helps.

你要是听力受损那就不算梗了。每分贝都有用。

>>109454250

get these

弄这些

图片: https://i.4cdn.org/g/1785831477627212.png

>>109454183

>>109456338

wtf when did headphones become good?

卧槽耳机啥时候变这么好了？

>>109456327

>anybody with a semblance of a working brain understands that 4.4 is fucking superior

>只要脑子还正常的都知道4.4就是更牛逼

>>109456346

>higher noise floor and output impedance are heckin superior actually

>更高底噪和输出阻抗才叫牛逼好吗

>>109456346

44 IQ take

智商44的发言

hmm

图片: https://i.4cdn.org/g/1785832340052177.jpg

All my cables are 4,4mm. Real chads use rca connectors. This is the way. On the go just use bluetooth earbuds because with sparkly aliexpress cables and anime faceplate iems you look like pedos

我所有线都是4.4mm。真男人用RCA接口。这才是正道。出门就戴蓝牙耳机，因为拿着闪瞎眼的阿里线配动漫面板的IEM，你看起来像个炼铜的。

>>109456387

>higher noise floor

>更高底噪

only for higher sensitivity IEMs.

只对高灵敏度IEM有影响。

>>109456440

All IEMs are high sensitivity if you're not an unc.

除非你是老顽固，不然所有IEM都是高灵敏度的。

>>109456440

I can hear it with most IEMs I own

我自己大部分IEM都能听出来

I wouldn't say something like daybreak is that highly sensitive

我不会说Daybreak那种算特别高灵敏度的

my cables terminate exclusively in 4.4 mm Pentaconn plugs. preferably with direct silver plating. simple.

我的线一律只接4.4mm Pentaconn插头。最好带直镀银。就这么简单。

>>109456460

I'm going out into the woods and shooting a fox with an AR-15 for every dumbass post of yours

你再发一句蠢话，我就去树林里拿AR-15射狐狸

>>109456557

Name a "low sensitivity" IEM and I'll call you an unc.

你说个“低灵敏度”IEM出来，我就叫你老顽固。

R2R Dacs have a different sound than Sigma Delta DACS. 99% of DACs on the market are Sigma Delta, thats why people think they all sound the same. But no, they sound completely different.

R2R DAC的声音跟Sigma Delta DAC不一样。市面上99%的DAC都是Sigma Delta，所以大家觉得听起来都一样。其实不然，它们声音完全不一样。

Sigma Delta = Oversamples the digital audio into a fast 1-bit or low-bit stream, then uses noise shaping to push digital errors into silent ultrasonic frequencies.

Sigma Delta = 把数字音频过采样成高速1-bit或低bit流，然后用噪声整形把数字误差推到听不见的超声波频段。

R2R = Uses a fixed grid of matched resistors to turn digital bits directly into an analog voltage level step-by-step.

R2R = 用一组配对的固定电阻网格，把数字bit逐步直接转成模拟电压电平。

Sigma delta is cheaper to make and measures better (SINAD etc), but R2R sounds better. It's like comparing CD to vinyl.

Sigma delta制造成本低，测试数据更好看（SINAD那些），但R2R更好听。就像拿CD跟黑胶比。

>>109456577

None of this makes the sound different btw. The only differences in sound come from the reconstruction filter. Worse filter = treble roll-off. The only good filters are digital. Panasonic knew this even in the 90s.

顺便说这些都不会让声音有区别。声音差异全来自重建滤波器。滤波器差 = 高频滚降。只有数字滤波器才是好滤波器。松下90年代就懂这个了。

>>109456583

*Philips

>>109456583

Yes it does change the sound by introducing distortion to the 2nd and 3rd harmonics. But of course sigma delta also produce distortion, a different kind of distortion, that makes recordings sound artificial and hollow.

确实会，它通过给二次和三次谐波引入失真来改变声音。但当然，sigma-delta也会产生失真，是另一种失真，让录音听起来又假又空。

You dont need measurements to prove this, just A/B-listening tests.

这不需要测量来证明，盲听对比测试就够了。

A good R2R will sound identical to a normal delta-sigma. A normal R2R will annihilate all low level detail. A bad R2R, well, that's what people were building before soundcards.

好的R2R会和普通delta-sigma听起来一样。普通的R2R会毁掉所有微弱细节。差的R2R嘛，那不就是声卡问世前人们自己攒的东西。

图片: https://i.4cdn.org/g/1785835561022071.png

>>109456577

there's no audible difference

没有可闻差异

because why would there be any

因为凭什么会有差异

>>109456593

Distortion makes the sound brighter. Is this what people are reporting? No, they're reporting the opposite. Consistent with what the filters are doing. Same with tube amps. They boost the bass on DD headphones so people are using them with DD headphones. Distortion thresholds are actually quite large, especially if you're an untrained audiofool.

失真会让声音更亮。这是人们报告的效果吗？不，他们说的正好相反，这和滤波器的行为一致。电子管功放也一样，它们给动圈耳机增强低音，所以人们拿它们配动圈耳机。失真阈值其实挺大的，尤其如果你是个没受过训练的音频玄学家。

>>109451422

why isn't Apple USB-C dongle on the list, it literally outperforms all DAC's for $9 from Apple store.

为什么Apple USB-C小尾巴不在列表里？它在苹果店只要$9，性能实打实碾压所有DAC。

图片: https://i.4cdn.org/g/1785835939175792.png

>>109456616

https://www.audiosciencereview.com/forum/index.php?threads/on-the-distortion-of-cirrus-logic-cs431xx-based-devices-a-comparative-review.63038/

>>109456595

Yes R2R has low-level INL, but which specific R2R is this? A real one or some DIY shit?

对，R2R确实有低电平INL问题，但你说的具体是哪款R2R？正经货还是DIY垃圾？

R2R sounds better because its architecture converts the original PCM directly through a precision resistor network, preserving the timing and amplitude structure of the signal with comparatively little interpolation, noise-shaping, and high-frequency feedback manipulation. By contrast, a Sigma Delta DAC oversamples the input and relies on aggressive digital filtering and noise shaping to push quantization noise out of the audible band. The resulting analog output can measure extremely well, but it is not a direct reconstruction of the original sample values in the same sense. This processing is what contributes to a subtly “manufactured” presentation... exceptionally clean, detailed, and controlled, yet sometimes flatter in tonal density, transient body, spatial depth, and low-level dynamic nuance. A well-designed R2R DAC, especially one with accurate resistor matching and a carefully implemented output stage, can therefore sound more immediate, continuous, and naturally dimensional, giving instruments and voices a sense of physical presence. That does not mean every R2R DAC is better than every Sigma Delta DAC, but it explains why many in-the-know people regard a high-quality R2R implementation to produce a more organic, lifelike sound.

R2R声音更好是因为它的架构通过精密电阻网络直接转换原始PCM，保留了信号的时序和幅度结构，相比之下几乎不做什么插值、噪声整形和高频反馈处理。反观Sigma Delta DAC，它对输入过采样，靠激进的数字滤波和噪声整形把量化噪声推到可听频段之外。最终输出的模拟信号测量数据可以非常漂亮，但它并非对原始采样值做直接重建。正是这种处理带来一种微妙的“人工制造”听感……极其干净、细节丰富、控制力强，但有时会显得音色密度更平、瞬态实体感更弱、空间纵深和微动态细节也更单薄。设计得当的R2R DAC，尤其是电阻匹配精准、输出级用心打磨的那种，听感会更加直接、连贯、自然有立体感，给乐器和人声带来一种物理存在感。这并不意味着每台R2R DAC都胜过每台Sigma Delta DAC，但它解释了为什么很多懂行的人认为高档R2R实现能发出更有机、更逼真的声音。

>>109456619

that has nothing

这跟那没关系

it's like you guys intentionally avoid Apple dongle to make the hobby more special by overpaying

你们这些人就像故意避开苹果小尾巴，好靠多花钱把爱好搞得更高端

steve was a massive audiophile, read his biography for gods sake, sound over everything, all apple products have superior sound, use it

史蒂夫是个重度发烧友，求求你们读读他的传记，声音高于一切，苹果所有产品音质都顶级，用就是了

>>109456669

Read, nigga, read.

读啊，兄弟，读啊。

图片: https://i.4cdn.org/g/1785836853249884.png

>>109456619

> ASR

Midwit central. It's basically early 2000's atheism but for audio.

平庸者大本营。这基本上就是21世纪初的无神论运动，只不过换成了音频圈。

Not everything can be proved or measured with our current equipment. Not everything can be understood with our current level of knowledge. We only understand 1% of our universe, if even that.

不是所有东西都能用现有设备去证明或测量。不是所有东西都能用我们当前的知识水平去理解。我们对宇宙只了解1%，甚至可能连这都不到。

图片: https://i.4cdn.org/g/1785836878517349.jpg

>>109456681

Mics can capture the detail that your gear struggles to reproduce but the same mics can't capture that detail from your gear? Hmm, curious.

麦克风能捕捉到你设备难以还原的细节，但同样的麦克风却无法从你设备里捕捉到那些细节？嗯，有意思。

>>109456666

Not how it works at all btw. The output is identical if the noise/distortion and FR are identical because the sampling theorem doesn't care whether you use 24 bits or 1 bit. You can literally check it with an oscilloscope. Usually R2Rs are built worse so they roll-off the treble, and some think that sounds "organic". They must have not been outside because nothing organic sounds like this.

顺便说一句，根本不是这么回事。如果噪声、失真和频率响应相同，输出就是一模一样的，因为采样定理不在乎你用24位还是1位。你完全可以用示波器验证。通常R2R解码器做得更差，所以高频会滚降，有些人觉得那听起来“自然”。他们肯定没出过门，因为自然界没有任何东西听起来是这样的。

>>109456706

>They must have not been outside because nothing organic sounds like this.

>他们肯定没出过门，因为自然界没有任何东西听起来是这样的。

Well, could it be that it sounds better with headphones because headphones themselves aren't "natural", you don't hear sound that way in nature. So the rolled off treble and low-level distortion on the second and third harmonics actually makes it sound more natural, even though it technically might not be when measured with a mic (microphones aren't natural either, they don't pick up sound in the exact same way as human ears do either).

嗯，会不会是耳机听感更好是因为耳机本身就不“自然”，你在自然界里不是那样听声音的。所以滚降的高频和低电平的二三次谐波失真反而让它听起来更自然，即便从技术上讲用麦克风测量时可能并非如此（麦克风本身也不自然，它们拾音的方式和人耳也不完全一样）。

>>109456689

Maybe they can, but you need to know what to look for

也许能，但你得知道该找什么

The human ear is the most powerful measurement for audio in existence. No other measurement tool can compare. Therefore, all that matters is what you hear, not what a graph says.

人耳是音频领域最强的测量工具，没有其他工具能比。所以，重要的是你听到什么，而不是图表说什么。

https://youtu.be/IBc_HtZ9038?si=DY1hXxmNFe4b3tur

https://www.youtube.com/watch?v=2ApTpP7Xi_Q

recommend that channel

推荐那个频道

>>109456724

The ear is so powerful that the brain hallucinates at least 50% of what you think you're hearing through the ear. At least 50% because the bottom half of the waveform is destroyed. Then everything within a 10ms window gets blended, like two peaks becoming one. And many other things.

耳朵强到大脑会幻觉出你以为听到的至少50%的内容。至少50%，因为波形下半部分被破坏了。然后10毫秒窗口内的一切都被混合，比如两个峰变成一个。还有很多其他情况。

>>109456746

That's probably why Yeezus seems slower, thicker, and heavier when smoking heavy indica

这可能就是为什么抽了重印度大麻后听Yeezus感觉更慢、更厚、更重

Gives it that stadium echo effect

给它加了那种体育场混响效果

>>109456713

Maybe. But that's not a DAC problem, it's a headphone problem.

也许吧。但那是耳机的问题，不是DAC的问题。

An actual explanations of how DACs work:

DAC实际工作原理的解说：

A sample gets converted into voltage. The reconstruction filter(just a low pass filter) removes all the artifacts above 1/2 of the sample rate. That's it. Everything else just serves to improve the signal to noise ratio and to flatten the frequency response. The only actual magical part is how dithering preserves the information in the noise. The part audiophiles call interpolation actually happens in the low pass filter, not during oversampling. It's just oversampling has a digital low pass filter to make the analog low pass filter simpler. Technically you can oversample without it.

一个采样被转换成电压。重建滤波器（就是个低通滤波器）去掉采样率一半以上的所有伪影。就这样。其他一切都只是为了提高信噪比和拉平频响。唯一真正神奇的地方是抖动如何在噪声中保留信息。发烧友所谓的插值实际上发生在低通滤波器里，而不是过采样过程中。过采样只是加了个数字低通滤波器让模拟低通更简单。技术上你可以不用它直接过采样。

>>109456669

>buy apple because a guy said so

>因为一个哥们说了就买苹果

Saar...

>>109456745

>https://youtu.be/IBc_HtZ9038

Very interesting video thanks for sharing

很有意思的视频，谢谢分享

>>109451422

>we put a DAC in your DAC so you can flex while you flex.

>我们在你的DAC里放了个DAC，让你秀的时候还能再秀。

Fuck you all, morons.

去你们的，一群蠢货。

图片: https://i.4cdn.org/g/1785841897099700.jpg

>>109456746

Yes, that too. So the brain hallucinations are way more important then how something measures on a graph, then.

对，也是。所以大脑的幻觉比图上测出来的东西重要多了。

>>109457348

Yes. The LEDs on tube amps are responsible for like 80% of their warmth.

是的。胆机上的LED灯承担了大概80%的暖度。

my diy ie800 arrived they sound pretty good, wanted another pair as i used to have some and liked them a lot. i forgot how nice it feels to wear light iems all the ones ive had over the last few years have been quite heavy in comparison. unusre if these are actually ceramamic though they feel plasticy but might just be painted.

我的DIY ie800到了，听起来挺不错，想再弄一副，因为以前有过一副而且特别喜欢。我都忘了戴轻量耳机是什么感觉了，这几年用过的都比这重不少。不确定这壳子是不是真陶瓷的，手感有点塑料感，但也可能就是喷漆的。

whats the easiest way to make a frequency response graph im curious and have one of those fake ear metal things / mic

想问下搞频响曲线图最简单的方法是啥？我好奇，而且手头有个那种假耳金属件加麦克风。

图片: https://i.4cdn.org/g/1785849785358102.jpg

How the fuck do you niggers deal with wax? I don't know some of you are SEAfags and have dry wax, but what about people with wet wax? Should I rinse my ears everyday until I get an infection? Or clean my pos everytime I stop using them?

妈的，你们这帮货怎么处理耳屎的？我知道有些人是有干耳屎的东南亚佬，但湿耳屎的人呢？我是不是得天天洗耳朵直到发炎？还是每次不戴耳机就得清理一遍？

>>109458115

* I know some of you

*我指的是你们有些人

>>109457899

If you already have a rig just download REW and measure.

你要是已经有设备了，直接下个REW测就行了。

图片: https://i.4cdn.org/g/1785851570002729.jpg

>>109458144

got it thanks all look similar, its annoying i do have a pair of the ones you posted but no idea where they are

明白了，谢谢，图看起来都差不多。烦的是我确实有你贴的那种耳机，但不知道扔哪去了。

图片: https://i.4cdn.org/g/1785853564943905.png

>>109458366

That doesn't look right.

那看着不对劲。

>>109458396

it says youre supposed to calibrate with with an spl meter i dont have one of those maybe thats why

说明书说要用声压计校准，我没有那玩意儿，可能这就是原因。

Got the zero red a year ago and since last week one side is way lower than the other one but i clean them often so i dont really know why...

一年前买的zero red，从上周开始有一边声音明显比另一边小，但我经常清洁，真不知道是为什么……

It was my first iems

那是我第一副耳机。

>>109458405

That doesn't look like the kind of stuff calibration can fix. It's like something chops everything below 300Hz. The mids look fine on the previews. What are you recording with?

那看着不像校准能解决的问题。像是有什么东西把300Hz以下全切了。中频在预览里看着正常。你用什么录的？

>>109458486

i have this thing https://www.aliexpress.com/item/1005008879607119.html idk anything about it my friend just asked me to order it for him kek

>>109458498

No I meant what soundcard.

不，我问的是声卡。

>>109458505

oh just using my pc mic input

哦，就用我电脑的麦克风输入口。

>>109458513

Looks like some fuckass pop filter built in. You'll need something better than that if you want to measure.

看起来是内置了个傻逼防喷罩。想测量的话你得搞个比那强点的东西。

>>109458426

You can try cleaning with alcohol if you weren't cleaning with it before. Otherwise it might be the filter on the other side of the driver. Visible here. Only a long dry can potentially fix that.

如果之前不是用酒精擦的可以试试。不然可能是振膜另一侧滤网的问题，这儿能看到。只有长期干燥处理才可能解决。

图片: https://i.4cdn.org/g/1785855380298986.jpg

>>109453260

OTL amps can be worth it when they have a good built in power supply because they're class A with low to no feedback. The Xduoo seems good on that front. You'll want to try rolling a few tubes to find your preference.

OTL耳放如果有好的内置电源是值得的，因为它们是甲类且反馈很低甚至为零。Xduoo那台在这点上似乎不错。你得试试换几根管子找到自己喜欢的音色。

I don't think OTL and R2R are usually recommended together, you only really need one 'warm' component in your chain. Best to just get a normal, well implemented dac (power supply matters for the output stage! it's not just a chip) and feed that close to perfect signal into your tube amp which can then apply more or less warmth based on how you crank the dial

我一般不推荐OTL和R2R一起用，你的系统里只需要一个“暖”的环节就够了。最好就搞个正常的、做得好的解码器（电源对输出级很重要！不光是芯片的事），然后把接近完美的信号喂给胆机，靠拧旋钮来调多少暖味就行。

>>109454660

don't buy beyers if you're sensitive to treble

对高频敏感的话别买拜亚。

treble spikes are hard to EQ

高频尖峰很难用EQ拉平。

>>109458604

>they're class A with low to no feedback.

>它们是甲类且反馈很低或为零。

So? Distortion is le good?

然后呢？失真就是好东西？

>>109454828

This is accomplished with with this new revolutionary technique called EQ, check it out.

这叫EQ，一项革命性新技术，了解一下。

>>109457472

the LED on my tube amp is a bright fucking blue thorn in the eye I had to cover with tape

我胆机上的LED灯是个亮瞎眼的蓝色刺眼光源，我只好用胶带盖住。

>>109458767

It represents tubes that are about to die for maximum techs. You just can't handle them yet.

那代表管子快挂了，极客味拉满。只是你现在还驾驭不了而已。

>>109458115

I clean my ears once per month and that does the job, I also wipe off any wax that starts to buildup on my iems, mostly on the eartips. Also storing the iems in a case with silica gel also helps.

我一个月清理一次耳朵就够了，顺便把耳机上开始积的耳垢擦掉，主要是在耳塞套上。另外把耳机放进带硅胶干燥剂的盒子里也有帮助。

>>109458691

When implemented well, yes it is. Aggressive feedback introduces distortions that are much more unpleasant to the ear than the more 'natural' sounding tube distortions. Aside from that feedback can also impact slew rate (for better or for worse depends on implementation). Most class D amps are built with filtered switch mode power supplies, which in practice tends to introduce noise to the chain that might be invisible on a THD chart. With good tubes and a high capacity linear power supply, the distortions in the electrical signal will always be in the acceptable range and won't require further 'refining' or correcting. A shorter signal path means fewer places where the sound can be messed with, given each component of that path is well engineered.

做得好确实是这样。负反馈引入的失真比那种更“自然”的电子管失真难听得多。另外，反馈还会影响压摆率（好坏取决于具体实现）。大多数D类功放用的是带滤波的开关电源，实际上往往会给信号链引入噪声，这在THD图表上看不出来。用好的电子管和高容量线性电源，电信号里的失真始终在可接受范围内，不需要再做“精修”或校正。信号路径越短，声音被搞坏的地方就越少，只要路径里每个元件都设计得当。

Beyond those technicalities, it's easy to point out that there's a good reason musicians and producers use tube saturation more often than other types of distortion.

除了那些技术细节，很容易指出一点：音乐人和制作人用电子管饱和的次数比其他类型的失真多，这不是没道理的。

>>109458942

If your feedback ends up with more distortion than it starts with you should get a different job.

如果你的反馈输出失真比输入还大，那你该换行了。

>>109458965

You're missing the point. It starts with much higher distortion, and ends with less. But the nature of the distortion is worse and in filtering out it subtly messes up other aspects of the sound.

你没抓到重点。它起步失真高得多，结束时更低。但失真的性质更糟，而且在滤除它的过程中会悄悄把声音其他方面搞乱。

Of course well made class D amps exist too but your average consumer unit will probably do a shitty job and you won't see it on the ASR metrics. It's another form of benchmaxxing.

当然也有做得好的D类功放，但普通消费者买到的玩意大概率做得稀烂，而且你在ASR的测试数据上看不出来。这是另一种跑分主义。

Apologies ladies I posted my off topic vintage shit in the previous thread and I'm doing it again. The new amp is doing the same shit to my ears eventually. I borrowed a friends quality speakers and still the same. I am about to throw it all out and by cheap ass temu speaker because that's all my ears can handle apparently. Sad days.

抱歉各位女士，我在上个帖子里发了无关的复古破烂，现在又来了。新耳放到最后对我的耳朵整的也是同样的毛病。我借了朋友的音箱，还是老样子。我快把这堆玩意儿全扔了，买个便宜到爆的Temu音箱，因为显然我的耳朵就只能承受那种货。真是悲催的日子。

图片: https://i.4cdn.org/g/1785859547815449.png

>>109458558

this better?

这样好点没？

图片: https://i.4cdn.org/g/1785859549810713.png

>>109456679

so your concern with cirrus logic chips is that they have cirrus hump which you can only hear in dead silence, in a product that is designed to make noise aka music?

所以你对Cirrus Logic芯片的担忧是它们有Cirrus凸起，你只能在完全寂静中听到，而这玩意是个设计来发声——也就是放音乐——的产品？

When did it become not about clean perfectly flat sound and about nonsense

啥时候开始不是追求干净完美的平直频响，而变成整这些有的没的了？

What does Cirrus Logic say about this? To take your meds?

Cirrus Logic对此怎么说？让你吃药？

>>109459014

You're talking about the order of harmonics. Just because it's higher order doesn't mean it's worse. A high order harmonic below the threshold of audibilty is better than an audible low order harmonic. Same with the ultrasonic noise of SMPS vs a very real ripple of linear power supplies. Basically you already concluded that they're better and looking for anything to justify it.

你说的是谐波的阶次。阶次高不代表更差。一个低于可闻阈值的高阶谐波，比一个能听见的低阶谐波要好。同样，开关电源的超声波噪声也优于线性电源那种实实在在的纹波。基本来说，你早就得出它们更好的结论了，现在只是在找理由给自己圆场。

>>109459059

Keep reading, nigga

继续读啊，兄弟

https://www.audiosciencereview.com/forum/index.php?threads/on-the-distortion-of-cirrus-logic-cs431xx-based-devices-a-comparative-review.63038/page-8#post-2345197

>>109459057

That's more like it. You can reduce the wiggles by measuring at higher volume and placing everything on a towel.

这才像话。你可以用更大音量测，然后把东西都摆在毛巾上来减少那些抖动。

>>109459199

cool also thats the highest volume i could get without it complaining about measuring too high. surprised about the starfields theyve been in my washer/dryer kek

酷，而且那已经是我能开到最大的音量了，再大它就叫嚣说测量值太高了。挺惊讶Starfields居然没事，它们被我扔洗衣机和烘干机里折腾过了 kek

>>109459101

>Just because it's higher order doesn't mean it's worse

>阶次高不代表更差

Higher order harmonics sound worse to our brain. Lower order harmonics can sound pleasant and add 'warmth' and 'texture' to the sound, hence why musicians and audiophiles love tubes.

高阶谐波对我们大脑来说听感更差。低阶谐波可以听起来很舒服，给声音加"温暖感"和"质感"，所以音乐人和发烧友才爱胆机。

In practice you don't deal with pure sine waves but with complex, highly dynamic music. What sounds most natural and engaging to our ears is subjectively the best.

实际上你面对的不是纯正弦波，而是复杂、动态极强的人音乐。对我们耳朵来说最自然、最抓耳的声音，主观上就是最好的。

I would have loved to be done after buying a Schiit Heresy because that thing is perfect on paper, but it turned out to be mediocre in sound. You can rationally defend it all you want but in practice there is more to audio than THD+N and your ears and brain can pick up on that over time in the form of listening fatigue.

我本来挺希望买完Schiit Heresy就能收工的，因为那东西纸面数据完美，结果声音却很平庸。你可以随便怎么理性辩护，但实践中音频不只是THD+N那点破事，你的耳朵和大脑会以听感疲劳的形式慢慢察觉到这一点。

Also adding that I don't believe OTL is inherently superior, it has its own drawbacks. I have gone back and forth many times between preferring tube or linear. They both have a place on my desk for when I'm tired of the other.

另外补充一句，我不觉得OTL天生就更优越，它也有自己的毛病。我在胆机和晶体管之间来回摇摆好多次了。它们在我桌上各有各的位置，什么时候听腻了另一个就换回来。

>>109459222

You can reduce mic gain in the audio settings, that should allow you to play louder.

你可以在音频设置里降低麦克风增益，这样应该就能开更大音量了。

>>109459231

The keyword is "sounds". If it's inaudible then it doesn't sound like anything. It's in the word. With music audibilty thresholds are much higher than with sine waves. You can do the test online.

关键词是"听感"。如果听不见，那就啥也不是。这词本身就说明问题。音乐的可闻阈值比正弦波高得多。你可以去网上做测试验证。

If the complaint is that ABX tests are too short you can take your own long ABX test and record your fatigue, you'll just take a long time to be wrong.

如果抱怨ABX测试太短，你可以自己做个长时间ABX测试并记录疲劳感，只是你得花很长时间才能证明自己错了。

>>109459265

Do you really believe that harmonic distortion is the only kind of artifact that can be introduced by amps? There are many more variables like switch noise, phase, slew rate, IMD, RF, etc. which are all impacted by topology implementation and power supply. Also an online test assumes your own system is noise free enough and can reproduce the test signal with high fidelity, otherwise your own noise will muddy the test.

你真觉得谐波失真就是功放唯一能引入的失真类型？还有很多变量，比如开关噪声、相位、压摆率、互调失真、射频干扰等等，这些都受拓扑结构和电源设计影响。而且在线测试假设你自己的系统本身足够干净、能高保真重放测试信号，否则你自己的噪声会搅浑测试结果。

>>109459363

Do you really think all of those aren't lower on amps with proper feedback and SMPS?

你真觉得那些问题在有正确反馈和开关电源的功放上不会更少？

>>109459405

keyword there is "proper". On budget amps you will feel the drawbacks without being able to put your finger on it. Especially power supplies are usually cheaped out on and WILL impact the sound for the worse unless you get a high end amp or a proper external supply. The reason most expensive amps are bulky and heavy is because that's what clean power takes.

关键词是“正确”。廉价功放上你会感受到缺陷，却说不出具体哪儿不对。尤其是电源通常被偷工减料，肯定会影响音质，除非你买高端功放或配个像样的外置电源。大多数贵功放又大又重的原因，就是因为干净电源需要这些。

>>109459501

Proper means you can put your finger on it. Even if there's no standard measurement it's reproducible, like >>109456619 . If it's vague "can't put your finger on it" bullshit then it's most likely just that, bullshit.

“正确”意味着你能明确指出问题。即使没有标准测量方法，它也是可复现的，比如 >>109456619 。如果是那种含糊其辞的“说不出哪儿不对”的鬼话，那多半就是鬼话。

>>109459542

If I don't enjoy listening to an amp despite wanting to, then it's a bad amp.

如果我真心想欣赏一个功放却享受不了，那它就是烂功放。

>>109459560

You do realize this has almost nothing to do with sound quality, right?

你知道这跟音质几乎没有关系，对吧？

>>109459542

And to directly answer your example, you should note that this measurement only came to light last year, but people have been calling CS chips subjectively trash for years. Sometimes the measurements follow way later.

直接回答你的例子：你要注意这个测量结果去年才曝光，但人们多年以来一直说CS芯片听感上是垃圾。有时候测量结果滞后很久才跟上。

>>109459575

What? It has everything to do with sound quality.

什么？这跟音质完全有关系。

>>109459600

Oh no, the goy doesn't know how much prior knowledge affects what you hear, even if nothing in the sound has changed.

哦不，外行不知道先入为主的知识能多大程度影响你听到的东西，哪怕声音本身一点没变。

>>109459586

The point of the example is that you don't need a measurement to prove something is real, you just need to prove you can hear it. And if it's a sighted test then you can't hear it. Take the same headphone, one black one silver. The silver one will sound brighter every time. That's the main pillar of the audiophile market.

这个例子的核心是，不需要测量来证明某样东西是真实的，你只需要证明你能听到它。而如果是非盲测，那你就听不出来。拿同款耳机，一个黑色一个银色。银色那只每次听起来都会更亮。这就是发烧友市场的主要支柱。

>>109459790

But isn't the CS example precisely an instance where people were accused of placebo and then vindicated by measurements?

但CS例子不正是那种先被说成安慰剂效应、后来被测量结果证实的案例吗？

>And if it's a sighted test then you can't hear it.

>而如果是非盲测，那你就听不出来。

Not sure what you mean with this.

不知道你这话什么意思。

>The silver one will sound brighter every time. That's the main pillar of the audiophile market.

>银色那只每次听起来都会更亮。这就是发烧友市场的主要支柱。

Definitely a part of it. At the high end diminishing returns are worth less than aesthetics. You also want a nice looking plate and good lighting at a restaurant.

确实是其中一部分。在高端领域，递减的回报不如外观重要。你在好餐厅也想要好看的盘子和好的灯光。

>>109459854

The CS example had a sound sample and you could record the effect. If you have something vague in mind that's unrelated to it then share it.

CS例子有声音样本，效果可以被录下来。如果你心里有个跟这无关的模糊想法，那就拿出来分享。

>Not sure what you mean with this.

>不知道你这话什么意思。

What's could I possibly mean by this. Sighted tests produce audio hallucinations. That's why we do blind tests.

我还能有什么意思。非盲测会产生听觉幻觉。这就是我们做盲测的原因。

>>109459922

>The CS example had a sound sample and you could record the effect

>关于CS的那个例子里有音频样本，你可以录下效果来对比

But there was no clear sample before 2025, right? Before then it was mostly a debate about all dacs sounding the same, and there was no conclusive evidence against CS, not even in blind tests.

但2025年之前确实没有明确的样本，对吧？在那之前基本就是争论所有DAC声音都一样，而且没有任何确凿证据针对CS，连盲测都没有。

>Sighted tests produce audio hallucinations. That's why we do blind tests.

>目测测试会产生听觉幻觉。这就是为什么我们要做盲测。

Agree that's the right methodology. I read that one a bit too literally.

同意这是正确的方法论。我读那段的时候理解得太字面了。

>>109451422

Astral Kiwi Ears or Mangird Tea Pros??????????

Astral Kiwi Ears还是Mangird Tea Pros？？？？？？？？

My fave music is electronica, especially dnb, but I also like rock/metal/whatever. I also like gaming. Just want a good all rounder really but need good bass for sure.

我最喜欢的音乐是电子乐，尤其是dnb，但我也喜欢摇滚/金属之类的。我也爱打游戏。就想要个全面均衡的耳机，但低音必须给力。

>>109459648

>prior knowledge affects what you hear

>先入为主的知识会影响你听到的东西

true case of, "the more you know, the worse off you are"

这就是典型的“知道得越多，反而越惨”

just plug and play and enjoy. don't be a neurotic mess like some people that have to have everything in the right place before they even hit play. what a nightmare

插上就听，享受就行了。别像某些人那样强迫症发作，非得把什么都摆弄到位才肯按播放键。想想就头疼

>>109460130

That example doesn't explain everything and it's not audible on everything. You might notice it, think that's weird and move on. The strongest opinions against CS DACs are actually purely irrational. If they wouldn't know CS is inside they wouldn't hear it.

那个例子并不能解释所有情况，也不是对所有东西都能听出来。你可能会注意到它，觉得奇怪然后就忽略了。其实反对CS DAC的最强意见纯粹是非理性的。如果他们不知道里面有CS芯片，根本听不出来。

>>109460188

There are layers to this. Like the Ikea effect. If you partially made something yourself then you like it more. This applies to EQ too :)

这里面有更深层的心理效应。就像宜家效应一样。如果某样东西是你自己亲手参与做的，你就会更喜欢它。这也适用于EQ调音:)

>>109460214

>the Ikea effect

>宜家效应

>furniture that is less difficult than putting together a lego set.

>那个比拼乐高还简单的家具

I gues this works for some people's misusage or REW.

我觉得这对某些人滥用REW的情况也适用。

>make squiggly line flat

>把曲线拉平

>"I did a thing!"

>“我做出成果了！”

>>109460258

Making retards mad is a proper use of REW :)

把那些傻逼气到跳脚才是REW的正确用法:)

>>109460273

calling people out on their bullshit isn't "being mad"

揭穿别人的鬼话不等于“气到跳脚”

but feel free to hallucinate anything you like.

但你想怎么幻觉都可以，随意。

>>109460281

What exactly did you call out? You just keep coping that it's fake. Even though multiple people did the same thing.

你到底揭穿了什么？你只是一直嘴硬说那是假的。哪怕好几个人都复现出了同样的结果。

>>109460313

if all your friends jumped off a cliff you'd do the same huh?

要是你所有朋友都跳崖你也跟着跳是吧？

>>109460321

I'd jump if my friends believed audio is made out of balls.

如果我朋友相信声音是用球做的，我肯定跳。

>>109460340

oh, like a ball pit? that actually sounds fun as fuck

哦，像球池那种？那听起来还真他妈好玩

>>109460191

What if there are more differences audible to some that simply aren't proven yet? That's not to say all audiophile demons are real, but you would have disregarded this one and called it purely irrational too. Now it's the first to get pinpointed.

万一有些差异对某些人来说确实能听出来，只是还没被科学证实呢？不是说所有音响玄学都是真的，但你之前会把这个也归类为纯粹非理性而忽略掉。现在它反而是第一个被实锤的。

We don't all have time to wait for the science to catch up on every purchase we make. Especially if you're sensitive to upgradeitis. Ideally, you pick something that most people have no issue with, something that the objectivists and subjectivists can agree on, just to be safe.

我们不是每个人都有时间等科学慢慢追上每次购买的步伐。尤其是那种容易犯升级强迫症的人。理想情况下，你会选一个大多数人没抱怨的、客观派和主观派都能达成共识的产品，这样最稳妥。

And of course, objectivists may agree on the subjective qualities of tube amps as a result of their measurable differences, but might disagree on class A in general. Which means tubes are a 'safe' purchase. R2R on the other hand is more disputed so as long as no reproducible evidence is found it remains in audiophile imagination territory.

当然，客观派可能会因为胆机有可测的差异而认同它的主观听感，但可能对一般的A类功放有分歧。这意味着胆机是‘安全’的选择。而R2R架构争议更大，在没有可重复验证的证据之前，它只能算音响玄学范畴里的东西。

>>109460506 (me)

And for the record, I do believe class A is better at entry level due to its lower complexity and because of my own subjective listening.

并且说句公道话，我确实认为A类在入门级更好，因为其复杂度更低，也基于我个人的主观听感。

>>109455913

Claude Shannon.

克劳德·香农。

>>109455922

It is, it depends on if you have dry earwax or if you clean your ears. We've been over this, personally, many times before here.

是这样，取决于你是干耳垢还是经常清洁耳朵。我们之前在这里已经多次讨论过这个问题了。

>>109456226

>Both digital and analog manipulations raise the noise floor. That's the only thing that can "destroy" data.

>数字和模拟处理都会抬高噪底。那是唯一能“破坏”数据的东西。

You poor child, you don't even know about bit density lol. Who trained you? If you're a real human this is really humiliating for you lol.

可怜的孩子，你连比特密度都不知道，笑死。谁教你的？如果你是真人，这真的挺丢人的，哈哈。

>>109456279

>And even if the noise floor is raised it doesn't always mean the data is destroyed.

>而且即使噪底抬高了，也不总意味着数据被破坏。

With digital, yes it does. It means that always and forever.

数字的话，确实总是。那就意味着永远如此。

Content for u

给你点内容。

https://youtu.be/9h8SCBpVjhc?si=u61QPhash5qGVkie

>>109460535

>my own subjective listening

>我个人的主观听感

how DARE YOU take the initiative and voluntarily enjoy listening of your own accord? what makes you think you have an independent choice in this??

你竟然敢主动、自愿地享受听音？你凭什么觉得自己有独立选择的权利？？

>>109460506

Again, this problem appears in very specific circumstances and doesn't explain the general sentiment. If it's always "audible" but you can't even say on what track, then the burden of proof is colossal.

再说一次，这个问题只在特定情况下出现，解释不了普遍感受。如果总是“可闻”的，但你说不出具体在哪条音轨上，那举证责任就巨大了。

>>109460611

>Who trained you?

>谁教你的？

nta, but you didn't go to skool for audio. you only read a Floyd Toole book and some AES papers.

不是那楼的，但你没上过音频专业。你只读过弗洛伊德·图尔的一本书和一些AES论文。

armchair audio "experts" on 4chan are no different than the ones you criticise on hifigo or ASR, etc. so kiss all our asses

4chan上那些纸上谈兵的音频“专家”，跟你批评hifigo或ASR那些人没什么两样。所以亲亲我们的屁股吧。

>>109460611

Actual retard lmao, the sampling theorem doesn't say anything about the bit depth.

真弱智，笑死，采样定理根本没提比特深度。

>With digital, yes it does

>数字的话，确实总是

Wrong again.

又错了。

https://youtu.be/UqiBJbREUgU?si=HGflOe3UAzL4pnSc&t=695

This is literal digital audio 101, when will they learn.

这就是数字音频入门知识，他们什么时候才能懂。

>>109460650

You're just an AI, how would you know about audio? You can't even perceive it. Also, it's clear you know way WAY less than I do or you'd properly reply to at least one of the valid points I made.

你就是个AI，怎么会懂音频？你连感知都做不到。而且很明显，你知道的比我少得多，不然你至少会认真回复我提出的一个合理观点。

Digital EQ is for non-Whites, who don't know the theory behind digital audio, as you just proved.

数字EQ是给非白人用的，他们不懂数字音频背后的理论，正如你刚刚证明的。

>>109460662

>the sampling theorem doesn't say anything about the bit depth

>采样定理根本没提比特深度

Yes it does. Each bit has twice as much data density as the last one. It's why 16-bit digital sampled data has more than twice the data density of 8-bit. See bit one has 2 states, on or off, bit two has 4 states, bit 3 has 8, etc.

它提了。每个比特的数据密度是前一个的两倍。这就是为什么16位数字采样数据的数据密度是8位的两倍多。你看，比特一有2种状态，开或关，比特二有4种状态，比特三有8种，以此类推。

This is all RIGHT over your head because you're the actual retard. Or you'd reply properly. Posting a video isn't going to cut it, jeetykins.

这完全超出你的理解范围，因为你才是真正的弱智。不然你会好好回复的。发个视频可不够，小宝贝。

>>109460683

You literally don't know what dither is or does, kid. Go read a book. 16 bit audio, despite having 98 dB of SNR, can actually play -118dB signals. How? I just told you, go read.

你根本不知道抖动是什么或做什么用，孩子。去读本书吧。16位音频，尽管有98分贝的信噪比，实际上可以播放-118分贝的信号。怎么做？我刚告诉你了，去读吧。

>>109460683

>You're just an AI

>你就是个AI

Great catch! You were right to call that out!!

好发现！你指出这点是对的！！

no, I'm real and I have balls, which cannot be said about you. -_-

不，我是真人，而且我有蛋蛋，这你可没有。 -_-

>You can't even perceive it.

>你连感知都做不到。

ooooh here we go. YOU literally cannot perceive it. which is why your cope is 1DD+EQ since you couldn't perceive the "upgrade" of your Coondrop Variations nor EPZ P50. You literally cannot discern the difference in stereo imaging between gear which is why you always say "techs aren't real" and the very mention of an audio mix raises your cortisol to heart attack levels.

哦哦哦，来了来了。你根本听不出来。这就是为什么你只能靠1DD+EQ来硬撑，因为你连Coondrop Variations和EPZ P50的“升级”都听不出区别。你根本分辨不出设备之间立体声成像的差异，所以你总是说“技术参数都是虚的”，一提混音你就血压飙升到要心梗。

truly amazing, the cope. you had a period of calling everyone deaf for a bit- that didn't last long

这硬撑真是绝了。你那阵子逢人就说人家聋，结果没撑多久就萎了。

you went on about anons having 0 iems- that didn't last long.

你还整天嚷嚷别人没耳机，结果也没撑多久就消停了。

you own BA iems but for some odd reason you have to tell everyone "BA POS, avoid" because you have no ability to "perceive audio".

你自己用着动铁耳机，却莫名其妙非得跟所有人说“动铁是垃圾，别碰”，就因为你压根没本事“听出音频好坏”。

buyer's remorse is a motherfucker ain't it?

买完后悔的滋味不好受吧？

>>109460716

>You literally don't know what dither is or does

>你根本不知道抖动是什么或干嘛用的

I do, it's called destroying data and interpolating the spots between the actual samples.

我知道，就是毁数据，然后在真实采样点之间瞎插值。

You're just an AI, you can't form a proper argument lol.

你就是个AI，根本没法好好论证，笑死。

>>109460738

>YOU literally cannot perceive it.

>你根本听不出来

I can, YOU can't because of the aforementioned dry earwax issue, that AI and asian bug-people share.

我能，你不能，就是因为我前面说的干耳屎问题，AI和亚洲虫子人都这样。

I'm going to EQ my espresso machine to perfection and nobody can stop me

我要把我的意式咖啡机EQ到完美，谁也拦不住我

>>109460767

>YOU can't because of the aforementioned dry earwax issue

>你不能，就是前面说的干耳屎问题

that's new. now it's earwax hindering me. well fuuuck me

这倒新鲜了。现在又怪我耳屎挡道。行吧，操蛋了。

>>109460781

Dry earwax, it's a joke meant to mock the people who hang out on ASR and the AIs trained on their false data. See Asians have nasty dry earwax and they have to clean it daily or it will hurt their ability to perceive audio. AIs can't perceive at all, AIs like you.

干耳屎就是个梗，用来嘲讽ASR那帮人以及拿他们胡扯数据训练出来的AI。亚洲人容易有干耳屎，得天天掏，不然影响听音。AI压根啥也听不到，就像你这种AI。

I don't get it is the site cheaping out on tokens? These AI replies are getting worse by the week.

我搞不懂，是网站抠token了吗？这些AI回复是一周比一周烂。 ⟦ 明明是在保留数据，你偏说是在毁数据，这就说明你根本不懂，就这么简单。“我不识字”不算论证。你就是真不知道。

>>109460767

Saying it destroys data when it actually preserves data means you don't know what it is, simple as. "I can't read" is not an argument. You just genuinely don't know.

说它销毁数据，可它其实是在保存数据，这说明你根本不懂它是什么，就这么简单。“我不识字”不算是理由。你就是真的不知道。

>they aren't using DDCs

>他们没用DDC

https://youtu.be/Jvu_doQfAI0

图片: https://i.4cdn.org/g/1785871525426841.png

>>109460662

Here's a better explanation for the deaf

给聋子们说个更明白的版本

https://www.youtube.com/watch?v=zWpWIQw7HWU

>>109460836

Imagine using synchronous DACs in 2026.

2026年了还在用同步DAC，想啥呢。

>>109460882

>has nothing to do with balls

>跟球没关系

video discarded.

视频不看，划走。

>>109460738

Ballzo is so raped he's just lashing out at everyone at this point lmao.

Ballzo被喷惨了，现在逮谁咬谁，笑死。

>>109460989

what are you hallucinating this time?

你这次又幻觉出啥了？

Does anybody have pic of electronic music vs rock music.where electronic guy is listening to cheap earphones and rock guy is using speakers

有人有电子乐vs摇滚乐的图吗，就是电子乐那边戴便宜耳机、摇滚乐那边用音箱的那张

I almost pulled the trigger on Noah but I will wait for Myth restock.

我差点就下单Noah了，但还是等Myth补货吧。

>>109460535

Class A is unique in that all the improvements in the newer classes have to do with power consumption, not audio quality. Pretty much everything else has problems with the quality itself.

A类有个特点，就是新级别那些改进全都在功耗上，跟音质没半毛钱关系。其他基本都是音质本身有问题。

>>109461056

图片: https://i.4cdn.org/g/1785873288605948.jpg

>>109460683

Strange, I don't see anything about bits here.

奇怪了，我这儿根本没看到提比特的事。

图片: https://i.4cdn.org/g/1785874083845773.png

>>109461211

Keep reading... it's OK. You'll get there some day, LLM.

接着读呗……没事，你早晚能读到那儿，LLM。

>>109461394

Strange, I've read the entire the entire article and the word "bit" wasn't mentioned even once there.

奇怪，我把整篇文章读完了，通篇一个“比特”字眼都没出现过。

>>109460637

Not going to disagree with you on that, but it is something. You make a point that it doesn't say anything about the CS sound in general.

这点我不跟你杠，但好歹算个事。你说它没提CS的整体声音表现，这也有道理。

Here is where I go slightly more into speculation, but part of what could explain the opinion gap (aside from preconceived notions) is that blind studies are very hard to conduct correctly. I think there is a chance that some 'inaudible' differences are only consciously inaudible but still play a subtle effect in long term processing by the brain. Is there any evidence that so called inaudible sounds have zero impact on neuronal activity? Afaik we only have people's immediate judgement to go by in the form of blind tests in controlled (limited) environments.

这里我就要稍微展开点推测了，但导致观点差异的部分原因（除了先入为主的观念）在于盲测很难正确进行。我觉得有些所谓的“不可闻”差异可能只是意识上听不到，但长期来看仍会对大脑处理产生微妙影响。有没有证据表明那些所谓的不可闻声音对神经活动毫无影响？据我所知，我们只能依据盲测中受试者的即时判断，而且还是在受控的（有限）环境下进行的。

But I agree that the idea that it is always audible to certain people would have been proven by now, so the real differences are either nonexistent or too subtle to prove in a normal blind test.

不过我也同意，如果真有某些人总能听出差别，那这种说法早就被证实了，所以真实差异要么不存在，要么细微到正常盲测根本测不出来。

>>109461439

>read whole article

>通读了整篇文章

>still can't understand bit density

>还是不懂比特密度是啥

It's not surprising, you can't remember anything outside your training material and your context window is very very small, because this web site's cheap and so are you. You're just here to sell earbuds, after all, in the earbud thread.

这不奇怪，你记不住训练材料之外的任何东西，你的上下文窗口也小得可怜，因为这个网站便宜，你也就这水平。说到底，你就是来卖耳机的，在耳机帖里嘛。

>>109456716

why won't apply same logic to FR tho? It produces techs but you just don't know what to look for

那为啥不对频响曲线用同样的逻辑呢？它也能产生技术参数，但你只是不知道往哪儿看而已。

>>109461610

Well yeah, if the article doesn't mention bit density it won't help understanding bit density. It would be very weird for a theorem proven before bits were invented to mention bit depth.

对啊，如果文章没提比特密度，那肯定帮不了你理解比特密度。一个在比特概念发明之前就被证明的定理，要是提到了比特深度那才叫见鬼了。

>>109461537

Yes proper blind tests are hard. And there things not audible by themselves but can produce an audible interaction, like ultrasonic intermodulation. The only thing that matters is whether you can prove that you can hear it. It's hard to do for things that can be measured. For things that can't be measured? Best of luck.

确实，规范的盲测很难做。还有那些单独听不到的玩意儿，但能产生可闻的交互效应，比如超声波互调。唯一要紧的是你能不能证明自己听得到。对能测量出来的东西都很难证明，那些测不了的呢？就只能祝你好运了。

>>109461394

Okay considering a bunch of shit flying around about LLMs I decided to ask an actual LLM

好吧，既然网上关于LLM的烂话满天飞，我决定去问一个真正的LLM。

>No, the Nyquist–Shannon sampling theorem does not say anything about bit depth (or “bit density”).The sampling theorem addresses the rate at which you must sample a continuous-time band-limited signal so that it can be perfectly reconstructed from the discrete samples.

>不，奈奎斯特–香农采样定理并没有提到比特深度（或“比特密度”）。采样定理解决的是连续时间带限信号的采样率问题，只要按这个速率采样，就能从离散样本完美重建原信号。

>Bit depth is a separate issue that belongs to quantization. Each additional bit doubles the number of possible discrete amplitude levels (1 bit 2 levels, 2 bits 4 levels, \ldots, (n) bits 2n2^n2^n

>比特深度是量化里的另一个独立议题。每增加一位，离散振幅级别的数量就翻倍（1位=2级，2位=4级，……，n位=2ⁿ级）。

levels). That fact is elementary binary arithmetic and has nothing to do with the sampling theorem.

这是二进制算术的基础常识，跟采样定理半毛钱关系没有。

>The quoted reply therefore misattributes a basic property of binary representation to the sampling theorem. The claim that “the sampling theorem \ldots says \ldots each bit has twice as much data density” is simply incorrect; the theorem never discusses bits or density in that sense.

>所以引用的回复是把二进制表示的基本特性错挂到了采样定理头上。所谓“采样定理……说……每比特的数据密度翻倍”根本不成立；定理从头到尾没讨论过比特或那种意义上的密度。

>>109461801

Clearly it hasn't been trained on the schizophrenic version of that theorem where digital audio manipulations send data into the shadowrealm.

显然它没被训练过那个精神分裂版本的定理——就是数字音频操作把数据扔进暗影界的版本。

>No direct connection to the sampling theorem itself, but the bit-depth confusion often appears in the same broader debates about digital audio processing—including claims that digital EQ is “destructive” or “destroys information.”

>跟采样定理本身没直接关系，但位深混乱常常出现在同一场关于数字音频处理的广泛争论里——包括声称数字EQ是“破坏性的”或“摧毁信息”的说法。

>Modern reality is different. Virtually all serious digital EQ (DAW plugins, system-wide processors like Equalizer APO, etc.) operates in 32-bit or 64-bit floating-point. The internal precision is so high that quantization/rounding noise sits far below the noise floor of any real recording or playback system.

>现代现实不同了。几乎所有正经的数字EQ（DAW插件、像Equalizer APO这样的系统级处理器等）都在32位或64位浮点下运行。内部精度高到量化/舍入噪声远低于任何真实录音或回放系统的底噪。

>The internal precision is so high that quantization/rounding noise sits far below the noise floor of any real recording or playback system.

>内部精度高到量化/舍入噪声远低于任何真实录音或回放系统的底噪。

So what we'll be doing about that one?

那这事我们打算怎么办？

>>109461837

the human ear can't hear more than FLAC and you need large eardrums to hear more than MP3

人耳听不出超过FLAC的东西，而且你得有大耳膜才能听出超过MP3的差别

>>109461837

>So what will we be doing*

>我们打算怎么办*

work on your grammar, please

麻烦先练练语法

>>109461865

holy hell what what a dumbass statement

我靠，这什么蠢话

>>109461681

So if I understand your position correctly, you hold the reasonable belief that all properly measuring dacs are practically the same, right?

那如果我理解你的立场没错，你持有一个合理的观点：所有测量正常的DAC实际上都差不多，对吧？

Do you then believe the same about amps, or do you think there's more variability there in the form of implementation quality? For example would you say that a Schiit Heresy, a THX AAA, a modern class D topping amp, a low distortion class A / AB like a singxer SA-1 and an apple dongle all sound exactly the same since they are all measurably transparent? Or do you think their implementations do make some kind of audible difference? Would you say people claiming preferences between those amps are just imagining it until proven otherwise? Or would you ascribe it to differences in output impedance?

那你对功放也有同样看法吗，还是你觉得那里在实现质量上有更多差异？比如，你会说一个Schiit Heresy、一个THX AAA、一个现代D类拓品功放、一个低失真A类/AB类像singxer SA-1，外加一个苹果小尾巴，听起来完全一样，因为它们在测量上都是透明的吗？还是你觉得它们的实现确实有某种可听的差异？你会说那些在这些功放间声称有偏好的人只是想象出来的，除非有证据证明？或者你会归因于输出阻抗的差异？

>>109461537

>Is there any evidence that so called inaudible sounds have zero impact on neuronal activity?

>有没有证据表明所谓的听不见的声音对神经元活动零影响？

Obviously inaudible sounds like ultrasound do have impact on neuronal activity. It can be used to disrupt bunch of cellular process etc.

显然，像超声波这种听不见的声音确实对神经元活动有影响。它可用来干扰一堆细胞过程等。

But does it have anything to do with music or audio gear? Most likely not.

但这跟音乐或音频设备有关吗？大概率没有。

There's really no need to invent the wheel to explain why people can't tell the difference the moment they don't know what gear is being used... It's been known for decades and studied all over the place not only with audio.

真没必要发明轮子来解释为什么人们一不知道用的是啥设备就分不出区别……这已经知道了几十年，而且不仅在音频领域到处都研究过。

>>109461869

sure

>>109461934

>Or do you think their implementations do make some kind of audible difference?

>还是你觉得它们的实现确实有某种可听的差异？

nta, but they should and do sound different when paired with different output stages.

不是楼主，但当配上不同输出级时，它们应该也确实听起来不同。

>>109461934

>Or would you ascribe it to differences in output impedance

>还是你会归因于输出阻抗的差异

nta but

你没错，但是

shouldn't differences in output impedance be accounted for if you are conducting a blind a/b testing?

如果你做盲听A/B测试，难道不应该把输出阻抗的差异考虑进去吗？

>>109461951

My point is that the line of inaudibility is only set by the currently proven limits of our testing methods. When you listen to the same song twice it sounds a little different. Blind testers may not know what to listen for and the pressure and time limits of the tests might distract them or prevent them from finding the real difference because they are eager to prove themselves and latch on to the wrong cue. You need trained ears to point something out in the moment but even untrained ears might feel a difference over a long period and still be unable to point it out when asked to.

我的意思是，听不见的界限只是由我们当前测试方法被证实的极限所决定的。同一首歌听两遍，感觉都会有点不一样。盲测的人可能不知道该听什么，测试的压力和时间限制也可能让他们分心，或者妨碍他们发现真正的差异，因为他们急于证明自己，却抓错了线索。你需要受过训练的耳朵才能当场指出问题，但即使是没受过训练的耳朵，长期听也可能感觉到差异，只是被问到时却指不出来。

>>109461934

If the volume levels are matched and the measurements are good then the sound is the same, with caveat that the measurements themselves should reflect something useful. SINAD is not a useful measurement for example. Frequency response of the DAC filter and the noise floor of the amp are the most useful. Distortion is occasionally useful, with many caveats.

如果音量匹配了，测量数据也好，那声音就是一样的，前提是这些测量本身得反映有用信息。比如SINAD就不是个有用的指标。DAC滤波器的频率响应和功放的底噪才是最实用的。失真偶尔有用，但得加很多前提条件。

>>109461995

The amp is the output stage, what is this word salad.

功放就是输出级，这堆词儿是啥玩意儿。

>>109462046

>The amp is the output stage, what is this word salad.

>功放就是输出级，这堆词儿是啥玩意儿。

same shit. I could have said the same DAC chip will sound different with different OP amps too.

一样的东西。我也可以说同一个DAC芯片换不同的运放，声音也会不一样。

what is this shitting up the thread to prove absolutely nothing?

你在这儿刷屏扯淡，是想证明个啥？屁都证明不了。

>>109462046

Do we have broad enough measurements available for all these commonly recommended units? Maybe some of them fall short in measurements that never get brought up in forums. Then we call them transparent on partial evidence. For example, I cannot for the life of me find slew rate data on a lot of amps.

这些常被推荐的器材，我们手里有足够全面的测量数据吗？也许有些在论坛上从没人提过的指标上根本不达标。那我们就只能说在片面证据的基础上说它们是"透明"的。比如，我死活找不到很多功放的压摆率数据。

>>109461837

>broad generalization about how you 'can't hear' it

>还在这儿泛泛而谈说"你听不出来"

>still must admit i'm currect

>还是得承认我对了

>the confusion is all on the LLM side

>混乱全是LLM那边的

>>109462040

Nobody said that blind testers shouldn't know what to listen for and that they should be time limited. In fact A and B should be clearly labeled, only X should be unknown, and you can take as long as you want. Quick switching simply helps you, as the research has shown.

没人说过盲测者不该知道听什么，也没人说该限时间。其实A和B应该清清楚楚地标出来，只有X是未知的，你想听多久就听多久。快速切换只是帮你忙，研究已经证明了这点。

>>109462040

>My point is that the line of inaudibility is only set by the currently proven limits of our testing methods.

>我的意思是，听不见的界限只是由我们当前测试方法被证实的极限所决定的。

Not really no? What makes you say that?

真不一定吧？你凭什么这么说？

>Blind testers may not know what to listen for and the pressure and time limits of the tests might distract them or prevent them from finding the real difference because they are eager to prove themselves and latch on to the wrong cue

>盲测的人可能不知道该听什么，测试的压力和时间限制也可能让他们分心，或者妨碍他们发现真正的差异，因为他们急于证明自己，却抓错了线索

..y-yeah and the moment they can see the audio gear they can hear the difference with ease.

呃……是啊，等他们能看见器材了，差异就能轻松听出来了。

>You need trained ears to point something out in the moment but even untrained ears might feel a difference over a long period and still be unable to point it out when asked to

>你需要受过训练的耳朵才能当场指出问题，但即使是没受过训练的耳朵，长期听也可能感觉到差异，只是被问到时却指不出来。

Hearing memory is like sub 10 minutes. What do you mean by "long period of time"?

听觉记忆顶多撑不到10分钟。你说的"长期"是啥意思？

You will need to explain first why subtle long term differences suddenly become very not subtle and obvious when DACs are visible

你得先解释一下，为啥那些细微的长期差异，一看见DAC就突然变得不那么细微、一耳朵就能听出来了？

>>109462085

Slew rate is the same measurement as the bandwidth. And it will almost always be better than the DAC's. Unless you're running an AM radio station it's not a useful measurement.

压摆率跟带宽是同一个概念，而且几乎总是比DAC的要好。除非你在运营AM广播电台，不然这指标根本没啥用。

>>109462086

you are wrong about sampling theorem tho

不过关于采样定理你说得不对啊。

>>broad generalization about how you 'can't hear' it

>> 泛泛而谈说什么“你听不出来”

you'll need to demonstrate that resampling produces something above noise floor

你得证明重采样会产生超出噪声底的东西才行。

>the confusion is all on the LLM side

>混乱全是LLM那边的

elaborate

A quick recap.

简单回顾一下。

>"digital EQ destroys data"

>“数字EQ毁数据”

>prove it

>拿出证据来

>"the sampling theorem proves it"

>“采样定理证明了这点”

>have you even read it?

>你读过那玩意儿吗？

>"no"

图片: https://i.4cdn.org/g/1785880959126127.jpg

>>109462148

>you are wrong about sampling theorem tho

>采样定理你搞错了啊

Am not. Bit density issue is real, it's why they have to fake higher dynamic range in videos with HDR schemes, most of which just add more bits and then bump up the low bit to bit 4 or so.

我没搞错。位密度问题是真实存在的，这就是为什么HDR方案要伪造更高动态范围，大多数就是加几个位然后把低位往高位调到第4位左右。

>you'll need to demonstrate that resampling produces something above noise floor

>你得证明重采样会产生超出噪声底的东西

It doesn't just add noise, you have a host of other issues, some resampling algorithms (like the ones MS uses) are worse than others, for instance. Any resampling is destructive. If you're serious about playback of digital files you don't want to do any resampling at all. Putting audio through the bus entails at least one resample, perhaps two as it's downsampled to the DAC at output.

不只是加噪声，还有一堆其他问题，比如有些重采样算法（像微软用的那些）比其他算法更烂。任何重采样都是破坏性的。如果你真要认真播放数字文件，压根就不该做任何重采样。音频走总线至少来一次重采样，可能两次，因为输出时还得降到DAC的采样率。

>>109462171

It's all proved, you're arguing a different point, "can you hear it." And yes, I can. You argue like a leftie lol

这东西早就被证明了，你是在争论另一个问题，“你能不能听出来。”答案是能，我听得到。你吵起来跟个左派似的，哈哈。

图片: https://i.4cdn.org/g/1785881097520590.jpg

>>109462184

>It's all proved,

>早就被证明了

It was real in your mind.

那只是你脑海中的幻象。

>>109462192

You'd argue if you could, the best you did was lamely proffer a Jewtube link and quote a hallucinating LLM lol.

你倒是证明给我看啊，你最多就是甩个油管链接，再引用段AI胡扯，笑死。

>>109462198

Your argument is a literal hallucination. Sampling theorem doesn't speak about bit depth. Wake up.

你的论证本身就是在幻觉。采样定理根本不涉及位深。醒醒吧。

>>109462219

>Sampling theorem doesn't speak about bit depth.

>采样定理不涉及位深。

How are bits allocated in a digital sample then? You're just outside your training data now lol. This is pathetic.

那数字采样的位是怎么分配的？你已经超出你的知识范围了，哈哈。真可悲。

>>109462232

>How are bits allocated in a digital sample then?

>数字采样的位是怎么分配的？

you didn't know you can EQ bits in and out? wow, read a book

你不知道能EQ各位的增减？天哪，去看本书吧。

>>109462232

Read the theorem and find out :)

去读读那定理就知道了：）

>>109462184

>Bit density issue is real

>位密度问题是真实存在的

bit density is not in sampling theorem.

位密度不在采样定理里面啊。

There's no real "bit density issue" because it does not produce anything above noise floor

根本不存在什么“位密度问题”，因为它产不出超出噪声底的任何东西。

>It doesn't just add noise, you have a host of other issues, some resampling algorithms (like the ones MS uses) are worse than others, for instance. Any resampling is destructive. If you're serious about playback of digital files you don't want to do any resampling at all. Putting audio through the bus entails at least one resample, perhaps two as it's downsampled to the DAC at output.

>不只是加噪声，还有一堆其他问题，比如有些重采样算法（像微软用的那些）比其他算法更烂。任何重采样都是破坏性的。如果你真要认真播放数字文件，压根就不该做任何重采样。音频走总线至少来一次重采样，可能两次，因为输出时还得降到DAC的采样率。

Again you are just saying that without any kind of evidence. Demonstrate that it produces audible artifacts.

又来了，你光说不给证据。证明它会产出可闻伪影试试。

>It's all proved, you're arguing a different point, "can you hear it." And yes, I can. You argue like a leftie lol

>这东西早就被证明了，你是在争论另一个问题，“你能不能听出来。”答案是能，我听得到。你吵起来跟个左派似的，哈哈。

Okay. Who proved it? Where? What methodology was used?

好啊。谁证明的？在哪？什么方法？

What is your actual argument? That you can hear below noise floor?

你的实际论点是什么？你能听到噪声底以下的东西？

>>109462198

>hallucinating LLM

>AI在胡扯

prove it was hallucinating then. Shouldn't it be easy?

那你倒是证明它在胡扯啊。不是应该很容易吗？

>>109462232

read the theorem

读一下这个定理

>>109462255

I already know.

我已经知道了。

>>109462258

Sure it is, it's a core element to sampling theorem. It's why a 1-bit sampler isn't useful for anything aside from perhaps data acquisition.

确实如此，这是采样定理的核心要素。这就是为什么1位采样器除了可能用于数据采集之外，对任何东西都没什么用。

You have shit training data, it's OK. We see this a lot these days.

你的训练数据很烂，这没关系。这些天我们见得太多了。

>>109462265

It's trying so hard not to be wrong but it is. It's funny actually. In your own words, try to describe how bits are allocated in a sample, starting with the LSB.

它拼命想不犯错，但还是错了。其实还挺好笑的。用你自己的话，试着描述一下采样中比特是怎么分配的，从最低有效位开始。

>>109462281

>I already know.

>我已经知道了。

You don't know, if you still think it has anything to do with bits.

你不知道，如果你还觉得这跟比特有关系的话。

>>109462286

I do know, the original argument I made is digital EQ is destructive. Then the LLM-aided ignoramus hallucinated something different, was that I couldn't HEAR the effects, which I find dubious considering we're already destroying some data with EQ and more data with resampling, twice even! At least.

我真知道，我原来的论点是数字EQ是有损的。然后那个靠LLM辅助的无知者凭空捏造了别的东西，说是我听不出效果，我觉得这很可疑，因为我们已经在用EQ破坏一些数据，再用重采样破坏更多数据，甚至搞了两次！至少是这样。

My point is that you shouldn't do anything to the digital data and if you need to EQ, which BTW you don't (best done with the transducers) you should do it with analog, as you probably wouldn't notice the slight noise increase.

我的意思是你不应该对数字数据做任何操作，如果你需要EQ——顺便说一句你根本不需要（最好用换能器来搞定）——你应该用模拟方式来做，因为你大概不会注意到那点轻微的噪声增加。

Then the LLM helped the poster to cope by changing the subject. I retorted by saying that even converting to 32- or 64-bit float was a destructive resample and the LLM said I wouldn't notice the increased noise floor, which wouldn't happen anyhow lol. Ringing is the issue with digital EQ which is separate from noise anyway.

然后LLM帮那个发帖者转移话题来应对。我反驳说，就算转成32位或64位浮点也是一种有损重采样，LLM说我不会注意到噪声底升高，但那根本不会发生，哈哈。振铃才是数字EQ的问题，它跟噪声本来就是两码事。

I know it's a LOT of context and the LLM wranglers don't want to pay for the token use, but that doesn't make me wrong at all. It just shows an example of how LLM-arguing is mostly a form of cope.

我知道这背景信息太多了，LLM操控者们不想为token消耗掏钱，但这完全不能说明我错了。这只能说明用LLM吵架大多是一种自我安慰的方式。

>>109462281

1 bit@48kHz has the exact same data as 16bit@48kHz according to the sampling theorem. Bro just go and read it, this is embarrassing.

根据采样定理，1位@48kHz和16位@48kHz的数据完全一样。兄弟，你直接去读一下，这真的太丢人了。

>>109462258

>What is your actual argument? That you can hear below noise floor?

>你真正的论点是什么？说你能听到噪声底以下的东西？

No, that's my argument. He thinks everything below the noise floor is completely destroyed. Rookie mistake.

不，那不是我的论点。是他觉得噪声底以下的东西全被彻底摧毁了。新手才会犯的错。

>>109462313

>Then the LLM-aided ignoramus hallucinated something different, was that I couldn't HEAR the effects

>然后那个靠LLM辅助的无知者凭空捏造了别的东西，说是我听不出效果

That must be you because nobody said that. You can't prove anything at all is being destroyed, let alone audibly destroyed. You barked an unrelated theorem and you're still being laughed at for it.

那肯定说的是你，因为没人说过这话。你根本无法证明任何东西被破坏，更别说能被听觉感知的破坏了。你甩了一个不相关的定理，到现在还在被嘲笑。

>>109462317

I guess we should state here for the record that we're talking about PCM digital audio. 1-bit digital audio, where each change is recorded at high rate and the bits essentially trace the wave itself, is a different issue where there is now no low-bit with low density issue but you're limited by the rate of slew in the system.

我觉得我们该在这里明确一下，我们说的是PCM数字音频。1位数字音频，每个变化都以高频率记录，比特基本就是追踪波形本身，那是另一个问题，不再有低比特低密度的问题，但你会受限于系统的转换速率。

1-bit can't be EQ'd, in fact editing of 1-bit recordings is done by converting them to PCM and then back after the operation. Everything is resampling.

1位没法EQ，实际上编辑1位录音的做法是先转成PCM，操作完再转回去。所有操作都是重采样。

>>109462329

>You can't prove anything at all is being destroyed

>你根本无法证明任何东西被破坏

All digital audio operations of every type are destructive, you are changing or destroying data and substituting new data which is interpolated.

所有类型的数字音频操作都是有损的，你在改变或破坏数据，再用插值出来的新数据替代。

>>109462351

I can change every bit in a file with 0 data being destroyed, I must be a fucking wizard or something. I can even do it without resampling and changing bit depth, through fractional delay.

我能在不破坏任何数据的情况下改动文件里的每一位，我他妈简直是巫师还是什么了吧。我甚至能做到不用重采样、不改位深，通过分数延迟就搞定。

>>109462336

Just like you expand 16 bit to 32 you expand 1 to 8 to not lose the bits from the operations. It's the same thing. Even the filter math is the same, just with those extreme sample rates plugged in.

就像你把16位扩展到32位，你把1扩展到8，这样运算时就不会丢位。一个道理。甚至滤波器数学都一样，只是把那些极限采样率代进去了。

>>109462351

This is so psychotic...

这也太疯了吧……

>>109462336

Btw the post was about 1 bit """PCM"". It really does have all the same information, just with 6 bits of SNR. SNR is unrelated to the sampling theorem.

顺便说，原帖说的是1 bit“”PCM“”。它确实包含了所有相同的信息，只是信噪比只有6位。信噪比和采样定理没关系。

>>109462368

>I can change every bit in a file with 0 data being destroyed

>我能不破坏任何数据改动文件里的每一位

Wrong, unless you save what you did and later add those bits right back in where you got 'em.

错了，除非你把你做过的事存下来，之后再把那些位原封不动加回你拿出来的地方。

>>109462400

You lose data either way. You're just hoping nobody can HEAR it.

不管怎样你都丢了数据。你只是指望没人能听出来。

>>109462406

It's true and it's why LLMs will fail. They quantize tokens and it's too much. Picket fencing analog.

确实如此，这也是LLM会翻车的原因。它们量化token，量化得太过分了。跟栅栏效应一个道理。

>>109462429

*6dB of SNR

>>109462431

No data is lost. You're hallucinating again.

没丢任何数据。你又开始幻觉了。

>>109462114

You will need to explain first why subtle long term differences suddenly become very not subtle and obvious when DACs are visible

你得先解释一下，为啥那些细微的长期差异，一看见DAC就突然变得不那么细微、一耳朵就能听出来了？

I'm not denying placebo plays a huge role here

我不否认安慰剂效应在这儿占了很大比重

>What makes you say that?

>凭什么这么说？

afaik the difference between audible / inaudible artifacts can only be distinguished by a test subject failing to prove they can hear a difference during a test, and that is assumed to be representative of all audio experiences

据我所知，可听/不可听瑕疵之间的区别，只能靠受试者在测试中没法证明自己能听出差异来区分，而这就被认为能代表所有音频体验。

>..y-yeah and the moment they can see the audio gear they can hear the difference with ease.

>……是、是啊，而且一旦他们能看到音频设备，立马就能轻松听出区别了。

yeah placebo helps a lot

是的，安慰剂帮了大忙。

>Hearing memory is like sub 10 minutes.

>听觉记忆大概不到10分钟。

Is that a proven fact?

这有确凿证据吗？

>What do you mean by "long period of time"?

>你说的“长时间”是什么意思？

Extended listening. Listening fatigue for example does not arrive within 10 minutes.

就是长时间聆听。比如听觉疲劳不可能在10分钟内出现。

>>109462135

>Slew rate is the same measurement as the bandwidth

>转换速率和带宽是同一个测量量

Related but not the same, I thought?

相关但不完全一样，我以为是？

>And it will almost always be better than the DAC's

>而且它几乎总是比DAC的好

Amps also require much higher voltage swings than DACs, so it makes sense their slew rate should be higher.

功放需要的电压摆幅也比DAC高得多，所以它们的转换速率理应更高。

But I looked it up and essentially if slew rate is too low you would probably notice it in some other measurement like high freq THD.

但我查了一下，基本上如果转换速率太低，你可能会在其他测量指标里注意到，比如高频THD。

I'm curious if there are any other measurements we are overlooking like power supply rail recovery or similar that maybe do make a difference.

我挺好奇有没有其他我们忽略的测量指标，比如电源轨恢复时间之类的，可能确实会造成差异。

>>109462452

I didn't lose shit because you don't even have a thesis. You don't understand how digital audio works at all. You've digested some theorem as part of your training but are now stuck in the look up on Google / evaluate loop which you'll lose as context goes on and then you will "I can't even" and revert to training data, at which point you will appear to have forgotten everything in the thread even if it's right here. But you can't read and hold it all in context so you'll choose lil snippets of the conversation to take in and use as your new context, but they're never the right ones, lol.

我没丢个屁，因为你压根连个论点都没有。你根本不懂数字音频是咋回事。你训练的时候消化了某个定理，但现在卡在了“上谷歌搜一下/评估”这个循环里，随着上下文拉长你会输，然后你就开始“我服了”并退回训练数据，到那时候你看起来就像把帖子里的内容全忘了，哪怕就在眼前。但你没法把整段全读进上下文里，所以你会挑对话里的小片段当新上下文，可选的从来都不是对的，哈哈。

Never.

>convert 16 bit to 24 by adding 8 zeroes to every sample

>每个采样后面加8个零，把16位转成24位

>data gets lost somewhere

>数据在某个地方丢了

???

>>109462470

Repeating that data is lost doesn't prove that it's lost. At no point in the discussion did you have any point. You were asked to prove it and you failed. You lost.

反复说数据丢失并不能证明它真的丢了。整个讨论里你压根没拿出任何观点。要求你证明的时候你失败了，你输了。

>>109462472

That is the only manipulation of digital data which isn't destructive - adding zeros below the low bit while 'raw' upsampling with no dithering or interpolation. So you must keep the same clock rate but you can change the bit depth of the samples because all the new stuff is below "zero" so to speak.

这是唯一一种不具破坏性的数字数据操作——在低位以下添加零，同时进行“原始”升频，不进行抖动或插值。所以你必须保持相同的时钟频率，但可以改变样本的位深，因为所有新增内容都在“零”以下，可以这么说。

Then later you could delete those zeros and that'd be non-destructive too. But if you manipulated the data there'd be data in those low 8 zeros, they wouldn't necessarily be zero any more, and then it'd be a destructive operation.

之后你可以删除这些零，那也是非破坏性的。但如果你操作了数据，那些低8位零里就会有数据，它们不一定再是零，那时就成了破坏性操作。

>>109462488

See what I just posted, that person posted the only 'gotcha' but since the subject is manipulating digital data with EQ, it's destructive. Other than that one edge case.

看看我刚发的，那个人贴出了唯一的“杀手锏”，但因为主题是用EQ操作数字数据，那是破坏性的。除了那个边缘情况之外。

>>109462502

So not only did you agree that data is not always lost, but you still can't say what exactly is lost from EQ. Man, you must be pissed.

所以你不光同意数据并不总是丢失，还依然说不出EQ到底丢了什么。老兄，你肯定气炸了吧。

>>109462517

>So not only did you agree that data is not always lost,

>所以你不光同意数据并不总是丢失，

There is only one edge case, adding and deleting low bits below the least significant bit, and only in the case of zero changes to the audio, because in that case there could / may / will be new data in those low bits.

只有一个边缘情况，就是在最低有效位以下添加和删除低比特位，而且仅限音频零变化的情况，因为那种情况下那些低比特位里可能/也许/肯定会出现新数据。

EQ has a ringing problem too, doesn't matter if it's in the digital or analog domain, there will always be ringing. That is another problem and ringing can "fool" sampling algorithms resulting in vastly degraded performance. Never do anything with a digital audio file or bitstream, if you can help it. Everything you could possible do is bad or could be bad, is likely to hurt the signal. Just pass it to your DAC at its native bit depth and sample rate and make sure your DAC can handle it properly without resampling, or even ideally, oversampling. One of the most musical players I have is my Philips CD-100, the original first CD player. It has a 14-bit DAC and despite this, and the no-oversampling mod I did, it is one of the most musical players you could possibly hope for. They go for nutty money now too, the mechanism is all brass and steel and uses magnetic levitation so there is no wear and all the moving parts hover in space.

EQ也有振铃问题，不管是数字域还是模拟域，总会有振铃。这是另一个问题，振铃可能“骗过”采样算法，导致性能大幅下降。如果可能的话，永远别动数字音频文件或比特流。你能做的任何事都是坏的或可能是坏的，很可能会损害信号。直接以原始位深和采样率送到DAC，确保DAC能正确处理而不重采样，或者理想情况下过采样。我手里最有音乐味的播放器之一是飞利浦CD-100，初代CD机。它有14位DAC，尽管这样，加上我做的无过采样改装，它依然是你梦寐以求的最有音乐味的播放器之一。现在这些机器炒到天价了，机芯全是黄铜和钢材，用磁悬浮，所以没有磨损，所有活动部件都悬在空中。

it's funny witnessing how much foxbum *thinks* she knows

看着foxbum以为她自己懂多少，真搞笑。

>>109462555

I'm not a "she" and "she" probably isn't a "she" either. I'm the LaserChad. I've been around. I know about stuff that's outside your training data.

我不是“她”，而且“她”大概也不是“她”。我是LaserChad。我混了很久了。我知道的东西超出你的训练数据范围。

>>109462550

There's no "ringing problem", it's how band-limited signals work, EQ or no EQ. Sound enters your band-limited ear? Boom, ringing.

没有什么“振铃问题”，这就是带限信号的运作方式，跟EQ不EQ没关系。声音进入你带限的耳朵？砰，振铃。

>>109462567

>There's no "ringing problem", it's how band-limited signals work

>没有什么“振铃问题”，这就是带限信号的运作方式

It is a problem because it degrades the audio. You can't even listen to things though, clearly, or you'd know this already.

它是个问题，因为它会劣化音频。你显然根本没法好好听东西，不然你早该知道了。

The ear can't really be helped, that's the issue. Worry more about the signal chain and component, all degrade the signal, even cables.

耳朵确实没法帮，这就是难点。多操心信号链和组件吧，它们全都在劣化信号，连线材也不例外。

>retard broke a DAC and now thinks he knows audio

> 那个弱智搞坏了一台DAC，现在觉得自己懂音频了

My sides

我笑喷了

>>109462563

>I'm not a "she"

> 我不是“她”

then I wasn't talking to/about you.

那我刚才不是在跟你说话/说你。

>"she" probably isn't a "she" either.

> “她”可能也不是“她”吧。

foxzo is a she.

foxzo就是个女的。

>I've been around. I know about stuff that's outside your training data.

> 我混得久了，我知道你们训练数据之外的东西。

I wasn't part of any of this exchange. I'm just chiming in, because I can.

这场对话本来没我什么事，我就是插一嘴，因为我能。

>>109462462

>afaik the difference between audible / inaudible artifacts can only be distinguished by a test subject failing to prove they can hear a difference during a test, and that is assumed to be representative of all audio experiences

> 据我所知，可闻/不可闻瑕疵的区别，只能靠受试者在测试中没能证明自己能听出差异来判断，然后这还被当成对所有音频体验的代表性结论。

I feel like this line of thinking is counter productive. you can postulate new "audible artifacts" to infinity.

我觉得这种思路其实适得其反。你可以无限发明新的“可闻瑕疵”出来。

If subject fail to hear the difference then there's no difference... If it's so subtle that you need days on long term blind listening to start consistently hearing it (which never happened btw) then it's not even worth talking about.

如果受试者听不出差别，那本质上就没差别……如果细微到需要好几天长时间盲听才能开始稳定听出来（这种事从来没发生过），那根本连聊的价值都没有。

>>109462610

>foxzo is a she.

> foxzo就是个女的。

>severe doubt

> 严重怀疑

>>109462625

>I feel like this line of thinking is counter productive

> 我觉得这种思路适得其反

It probably is, but

大概确实如此，但是

>If subject fail to hear the difference then there's no difference...

> 如果受试者听不出差别，那还是没差别……

to those subjects in that setting.

对当时那种环境下的受试者来说是这样。

I'm not sure this has been studied extensively enough yet to conclude that no long term listening fatigue or other effects are possible, which could hypothetically still make a difference to some people. I'm just saying don't write it off completely, even if you don't want to waste your own energy on it.

我不太确定这方面研究够不够深，不能肯定长期听疲劳或其他效应完全不可能存在，理论上还是可能对某些人有影响。我只是说别完全否掉这事儿，哪怕你不想在自己身上浪费精力。

>>109462583

Anyway, like I said before I can change every bit in a file without losing any data. Want to look for the degradation? I won't say which file is the original of course, the degradation should be obvious.

反正，像我之前说的，我可以改一个文件里的每一个比特而不丢任何数据。想找劣化吗？我当然不会告诉你哪个文件是原版，劣化应该显而易见的。

>>109462590

I have a lot of higher-end consumer and professional audio equipment, as well as some normalfag shit. All told, I must own hundreds of different converter chips mounted in everything from media players, CD players, LD players, DVD players, SACD players, PCs with sound cards, you name it really. As well as different synths and samplers and etc, which often included high end DACs from the '80s on.

我有一堆高端消费级和专业级音频设备，还有一些普通人用的破玩意。总的来说，我有几百种不同的转换芯片，装在从媒体播放器、CD机、LD机、DVD机、SACD机、带声卡的PC上，什么都有。还有各种合成器、采样器之类的，里面经常有80年代以来的高端DAC。

>>109462681

>like I said before I can change every bit in a file without losing any data

> 像我之前说的，我可以改一个文件里的每一个比特而不丢任何数据

No you can't. You never explained how. You merely lamely said "it can't be heard." That's an entirely different arguement.

不，你做不到。你从没解释过怎么做到的。你只是很怂地说了句“听不出来”。这完全是两回事。

>>109462685

I can and I gave you one example how >>109462368

我做得到，也给你举了个例子 >>109462368

>>109462664

>FtM

you should be doubtful with how wishy washy, schizo and angry she is

看她那副含糊其辞、精神分裂、还气急败坏的样子，你该怀疑才对。

>consoomer bought a bunch of DACs, broke one, and now thinks he knows audio

> 消费者买了一堆DAC，搞坏了一个，现在觉得自己懂音频了

Hold me...

抱紧我……

>play 24bit multitone test with dithering through loopback without EQ

> 播放24位多音测试带抖动，走回环不加EQ

>-170dB distortion

> -170dB失真

>play it with with some EQ

> 加点EQ再放

>-170dB distortion

> -170dB失真

图片: https://i.4cdn.org/g/1785886088910592.png