Hacker News Reader

Comment by earthnail | Hacker News Reader

Oh my, 192kHz/24bit. One would expect Apple would not need snake oil to sell their product.

To those not so familiar with audio DSP: 192kHz and 24bit make perfect sense if you intend to mangle the audio a LOT afterwards. At 192kHz, you can pitch shift your audio down by two octaves and still have frequency content in the entire CD spectrum. Very useful, and common, for movie effects. And 24bit, or better even 32bit, basically eliminates any floating point errors when you have a DSP chain with lots of multiplications.

But for playback, you need around 32-40kHz sample rate and 10 or 11bits. Even 16bit is already overkill.

View on HN

192kHz/24bit is definitely overkill but I wouldn't call 16/44 overkill; I'd say it's just about perfect and the engineers who decided on that standard when the CD came to market knew what they were doing.

The frequency range of human hearing does cover all the way to 20 kHz, and maybe a tiny bit beyond especially for kids and teenagers. That says 40kHz minimum sampling rate, 44.1 is enough to cover the full range, give a tiny bit for the antialiasing filter rolloff band, and the exact number was chosen IIRC because it fit well into repurposing existing 80s video recorders for digital audio.

The dynamic range of human hearing from threshold of audibility to threshold of pain is slightly wider than the theoretical 96 dB of 16-bit audio, but with dithering the effective dynamic range can be > 105 dB. Of course most popular music only uses a tiny bit of that range, but movies and even classical full-orchestral music can use 60 dB and beyond, for which 10 bits wouldn't be adequate.

192kHz is overkill of the overkill, but 10bit, seriously?! That is 1024 levels for the whole dynamic range, you are cutting it down to its 64th fraction man! With classical music where it is not filled up to the level of clipping but silent parts contain plenty of details you really must be joking with 10 bits or perhaps rely on listening experience of rap with through the roof bass and inarticulate jabber, there even 6bits may be too much, I give you that.

Also the standard 44.1kHz is good to have so the waves could be followed with little deviation and for the sake of reproduction. 192kHz is really snake oil but please do not jump over the horse to the other side in the frenzy of compensating.

Please study https://www.headphonesty.com/2019/07/sample-rate-bit-depth-b... or others.

> With classical music where it is not filled up to the level of clipping but silent parts contain plenty of details […]

And just because the music is silent doesn't mean the recording is: there's one album I have where there is a short rest in the musical work, but they recorded in a church in the country/rural area and seem to have have left the window(s) open, so during the musical stop you can hear birds chirping in the recording (if you have decent headphones).

On Pat Metheny's Travels album, there's a live recording of Farmer's Trust. It has birdsong added in by one of his band members (the name escapes me)

For a long time the version in Spotify didn't have these.

In a similar vein, in Vince Guaraldi's Live from Grace Cathedral album, you can hear coughing from the congregants in the background of some songs, notably Theme to Grace. Crush the dynamic range and it disappears

Even in "loud" music 10 or 12 bit is definitely noticeable. Many modern samplers include a legacy mode meant to mimic the grungy sound of old 10 or 12 bit sampling hardware. It might not be noticeable to non-musicians, but anyone who makes music will hear the difference. Even if you turn off other features that are usually included in tandem, like dac emulators.

That's when it's used as a "bit crusher" effect by converting to 10/12 bit samples by simple truncation, rather than dithering (randomized rounding up or down to the nearest quantization value, with P(round_up) = the relative closeness of the sample to the upper quantization value).

With dithering, the only difference a smaller bit depth makes is a higher noise floor, with the noise being just background white noise (like tape hiss). With most pop music a 10 bit depth with dithering would likely not be audibly worse, since pop music is usually compressed into like the top 10dB of the dynamic range.

>perhaps rely on listening experience of rap with through the roof bass and inarticulate jabber, there even 6bits may be too much, I give you that.

I think you made a great recap on why this is good and especially good for classical music listeners. But I can't fathom why so many music lovers online are only able to communicate about their love of music while trashing a specific genre. I think more people should be exposed to high quality audio and know how to appreciate it, theorizing to the ignorant that their tastes are wrong because their preferred music is bad probably doesn't move the needle in that direction, and IMO, reflects a different ignorance.

It wasn't about the genre or taste at all but about people accustomed to only less demanding settings drawing wrong conclusion based on that, projecting it on else. The situation looked like that.

Have you personally tested this in a blind study? I have, and the higher quality made a noticeable difference and I was able to identify it and prefer it across all test cases. The others I tested with did not notice any difference. It could be one of those cilantro things idk. But other accounts are in line with my experience: https://audiophilereview.com/audiophile-music/yes-19224-file...

Is 192/24 discernible to most people with mostly common listening equipment? No.

Should we be complaining that an audio streaming service is offering the highest quality audio? No. Especially because they're not charging a premium for it.

Apple was one of the last streaming providers to adopt lossless audio. Even then, the "high res" part is optional. You can disable it and stream normal 44.1/16

Reply on news.ycombinator.com