3 ms·

> like the idea that you could get a significant quality improvement sampling audio at 96k as opposed to 44.1k or 48k. There can be real quality improvements,

by dahart 2y ago

> like the idea that you could get a significant quality improvement sampling audio at 96k as opposed to 44.1k or 48k. There can be real quality improvements, if you’re editing, recording, mastering, producing audio. If the thing you sample isn’t the same as the thing that’s going to be played back, then a higher sampling rate is sometimes called for. This is true in imaging too. You don’t need HDR or raw formats or excessive resolution if you only show the same thing you captured, but sometimes you do if you’re going to modify it. Some of the lore around high sampling rates is when people make a rule of thumb for recording and apply it always without thinking about the specifics of a situation. That’s easy to do, but you also have to examine the cost of using a high sampling rate. Does it cost a lot or hurt? If you’re running short on storage space, maybe, but usually no, it’s usually a tradeoff with very low downsides. Sales people and corporate marketing absolutely do take advantage of this, for sure, and turn what are legitimate tradeoffs for some professionals into specious needs for mass market consumers.

PaulHoule 2y ago
If you want to do physics-based sound transformations which amount to applying partial or ordinary differential equations you have to sample at a rate 3x or more than the Shannon rate because otherwise you encounter instabilities. Similarly you can't really hear more than 16 bits worth of sample depth if you got your levels perfect but in the real word recording in the studio and field a few more bits of headroom make a big difference. Thus I'd expect people to record at a high sample rate but I think there is no practical delivery rate. I'm somewhat agnostic as to whether or not people can hear anything atr all past 44.1kHz, particularly because if you consider how accurately people can point out objects in a real physical soundfield (e.g. like a blind man shooting a gun) and you believe phase accuracy is the key, you'd make the case that people can hear timing differences that couldn't be captured at 44.1khz. But no "surround sound" technology comes close to replicating how accurately you can spot things in real life and I don't believe practically any commercial recordings past 44.1khz are noticeably better on any real equipment.
- dahart 2y ago
  Yep for playback, I agree, you don’t need more than 16 bits/sample or more than 44.1k playback rate. The vast majority of humans can’t hear beyond that, at least through the air… I have heard that it can go a lot higher via direct bone contact, but that’s not relevant to playing back music anyway. I just wanted to help clarify that there are real reasons to “sample” (record) something at 96khz or higher, or at 24 bits/channel/sample, but that the reasons easily get lost, and sometimes artificially or intentionally misrepresented for marketing purposes. I kinda relate this to sporting equipment or cars or other things people buy. Like, nearly everyone on a road bike is riding and wearing race gear, which is uncomfortable, while very few people are actually training or racing. Or, people often buy cars or computers that are far faster or more powerful than they need. Those things get rationalized similarly, with claims that the extra is necessary and makes a difference, or fears that it might be needed someday, and little reflection on the downsides.

by dahart 2y ago

PaulHoule 2y ago

If you want to do physics-based sound transformations which amount to applying partial or ordinary differential equations you have to sample at a rate 3x or more than the Shannon rate because otherwise you encounter instabilities. Similarly you can't really hear more than 16 bits worth of sample depth if you got your levels perfect but in the real word recording in the studio and field a few more bits of headroom make a big difference. Thus I'd expect people to record at a high sample rate but I think there is no practical delivery rate. I'm somewhat agnostic as to whether or not people can hear anything atr all past 44.1kHz, particularly because if you consider how accurately people can point out objects in a real physical soundfield (e.g. like a blind man shooting a gun) and you believe phase accuracy is the key, you'd make the case that people can hear timing differences that couldn't be captured at 44.1khz. But no "surround sound" technology comes close to replicating how accurately you can spot things in real life and I don't believe practically any commercial recordings past 44.1khz are noticeably better on any real equipment.

dahart 2y ago
Yep for playback, I agree, you don’t need more than 16 bits/sample or more than 44.1k playback rate. The vast majority of humans can’t hear beyond that, at least through the air… I have heard that it can go a lot higher via direct bone contact, but that’s not relevant to playing back music anyway. I just wanted to help clarify that there are real reasons to “sample” (record) something at 96khz or higher, or at 24 bits/channel/sample, but that the reasons easily get lost, and sometimes artificially or intentionally misrepresented for marketing purposes. I kinda relate this to sporting equipment or cars or other things people buy. Like, nearly everyone on a road bike is riding and wearing race gear, which is uncomfortable, while very few people are actually training or racing. Or, people often buy cars or computers that are far faster or more powerful than they need. Those things get rationalized similarly, with claims that the extra is necessary and makes a difference, or fears that it might be needed someday, and little reflection on the downsides.