Am I understanding basic audio science right?

[MRC01]

One way to partially resolve this would be to have the original live source and compare it to the recorded version. Some sources, such as my wife's voice are readily available (to me), and the human ear is highly tuned by eons of evolution to sense subtle errors in the voice. ...

This sounds like the notion that natural acoustic music is "the absolute reference" for audio. This doesn't work as well in practice as it sounds in principle, because it's actually a moving target. Your wife's voice (or any musical instrument) sounds quite different from 1 foot away as it does from 10 feet away. And it sounds different in every room of your house, and sounds different also depending on where you are relative to the source of the sound in that room.

All that said, I think the idea is still somewhat useful as long as one is willing to be a flexible listener. I know what instruments sound like from near to far, on stage playing with the musicians to sitting several rows back in the audience. And what they sound like in rooms of different sizes and shapes. These are all quite different, yet if the sound of the recording closely resembles any of these, I'm OK with that, will call it a "great" recording, and find the illusion convincing. However, many recordings (or audio systems) color the sound in ways that don't sound like any of these realistic variations. To me, these are not "high fidelity" in the true sense of the word.

 

[Wes]

This sounds like the notion that your wife's voice is "the absolute reference" for audio. This doesn't work as well in practice as it sounds in principle, because...

 

[redboat77]

This sounds like the notion that natural acoustic music is "the absolute reference" for audio. This doesn't work as well in practice as it sounds in principle, because it's actually a moving target. Your wife's voice (or any musical instrument) sounds quite different from 1 foot away as it does from 10 feet away. And it sounds different in every room of your house, and sounds different also depending on where you are relative to the source of the sound in that room.

All that said, I think the idea is still somewhat useful as long as one is willing to be a flexible listener. I know what instruments sound like from near to far, on stage playing with the musicians to sitting several rows back in the audience. And what they sound like in rooms of different sizes and shapes. These are all quite different, yet if the sound of the recording closely resembles any of these, I'm OK with that, will call it a "great" recording, and find the illusion convincing. However, many recordings (or audio systems) color the sound in ways that don't sound like any of these realistic variations. To me, these are not "high fidelity" in the true sense of the word.

Not an "absolute reference" but a practical and useful one. Any available sounds could be used for direct comparison: voices, the sounds of steps approaching, water running, etc. This is a practical suggestion - and is not intended to be an end all - but rather an improvement on the current widespread practice of judging systems based on how highly processed music recordings sound with no reference for comparison. Natural sounds are available for comparison and our ears and brains evolved to hear them to aid in our survival, and hence would likely be quite sensitive to poor reproduction.

 

[luft262]

Great question and thank you for opening up the debate! You'll get a lot of good answers here. I read some of them, but not all so sorry if others have already answered your question but here are my two bits...

1. Is it worth paying for 24 bit audio files with 48khz or higher, as opposed to CD quality tracks with 16 bit at 44khz?

A) In my humble opinion, yes, but barely. Let me explain. I haven't done any single or double blind tests of music files when I'm not in my listen/reading chair with my good dac, amp, and headphones. But I don't think I would notice any difference between Redbook CD (44.1 kHz and 16 bits) vs Hi-Res audio when using wireless earbuds at the gym, in my car, or even when using noise-canceling headphones while cooking. I don't think I could even hear a difference with my 5.1 home theater system with two subs and tower fronts. If you're listening in a car, while running, with lower-quality gear or in noisy environments I think you are unlikely to hear a difference. I have, however, done single-blind tests of Hi-Res audio and lossless CD quality vs 320mp3 and I could hear a difference in a quiet environment, when I was concentrating, and using my best amp/dac/headphones. I haven't done a test between Hi-Res and CD, because the streaming service I use, Qobuz, charges the same for Hi-Res and CD quality. So I was comparing Qobuz to YouTube Music, but I imagine any differences between CD and Hi-Res would be minimal even in the best of circumstances.

2. How good does my DAC need to be so that I can’t hear the difference with a better DAC?

A) There are various opinions about what is an audibly transparent bit depth, but Amir of ASR generally says 120dB or 20 bits his his level for audible transparency. If you are listening to MP3/AAC 320 kbs or something like that you probably don't need the full 20 bits, but it's nice to know its there. Once again it depends on your listening environment (noisy vs quiet), your equipment (expensive and well tested headphones vs cheap speakers), and how much you care and want to pay attention to the music you're listening to. If you mostly have music playing in the background while your playing a video game or driving, for example, you probably won't notice the same difference you would if you're sitting in a quiet room and focusing on the music with good speakers or open-backed headphones. You can get audibly transparent DACs and AMPS (ex. the JDS Atom DAC and AMP) for around $100, so you don't have to pay a lot of money to get to that 20 bit/120 dB threshold of audible transparency. More expensive DACS aren't always about audio quality. They offer more inputs/outputs, better build quality (to show off and for better user experience), and they support more file formats (think MQA for better or worse...lol).

3. Do I even need CD quality or is 320Kbs good enough?

A) IMHO for listening in noisy or distractive environments ex. driving, exercising, multitasking, or any other distracting or noisy environment AAC/MP3 at 320 kbs is fine. If you're critically listening in a controlled environment with good equipment getting to a lossless audio file/source such as CD or FLAC at 44.1 kHz and 16 bits will be the first significant step up in quality. Going from there to Hi-Res will be a very small difference, but your mileage may vary.

Overall, I think you should ask yourself how do you plan to listen to music. If you are, in fact, like many of the people on this forum and you plan to sit in a controlled environment and listen to music I think Hi-Res music isn't very expensive and may increase your listening enjoyment. The cost difference between Lossless CD quality and Hi-Res via streaming services isn't much. If you're trying to decide between say Spotify HiFi, which is CD quality lossless or Qobuz/Tidal, which offer Hi-Res audio, think about the price difference and other features you may care about between each service. The other tradeoffs may be worth more or less to you than Hi-Res audio. If you're mostly listening on the go and whatnot I would just stick with MP3/AAC 320.

But that's all just my opinion.

Thanks!

 

[Kal Rubinson]

I haven't done any single or double blind tests of music files when I'm not in my listen/reading chair with my good dac, amp, and headphones.

I don't think I could even hear a difference with my 5.1 home theater system with two subs and tower fronts.

Curious about this. Why do you think you can hear a difference in the first situation and not in the second? Is it headphones vs. speakers? Is it "my good dac, amp" vs. the equipment in your "5.1 home theater system ?" General context?

 

[MRC01]

Regarding high-res, I consider 44-16 to be musically transparent, but not perceptually transparent. By this I mean: certain sounds (jangling keys, square waves, and some others) when recorded in high res (say, 96-24) and then downsampled to 44-16, can be differentiated in DBT. So there is a small percentage of people (well trained listeners with good hearing) who can hear the limitations of 44-16. The differences are small and subtle, but audible to some people. But we don't listen to jangling keys, square waves, or other signals carefully devised to test the limits of the digital format. We listen to music. And most music doesn't challenge the limits of 44-16. Furthermore, 44-16 is almost perceptually transparent, you only a bit more data to achieve full perceptual transparency, 48-24 is probably sufficient.

That said, when it comes to music, if it's a high quality well engineered recording, I want it in its original native format. With a lot of modern classical music, that is high res. That way I'm avoiding resampling or other unnecessary processing. If that original format is 44-16, that's fine too. When it comes to anything else: recordings that are vintage (analog), or not well engineered (some classical, and pretty much all rock/pop etc.), high res is unnecessary. In fact, more often than not the high res version is worse than the original, as they've reprocessed it squashing the life out of the music.

 

[luft262]

Curious about this. Why do you think you can hear a difference in the first situation and not in the second? Is it headphones vs. speakers? Is it "my good dac, amp" vs. the equipment in your "5.1 home theater system ?" General context?

Well, that's a good question. Not all of my audio equipment has been reviewed on ASR so I can't be sure it can accurately produce Hi-Res music, but my primary listening setup has been reviewed. It is a JDS Atom DAC, JDS Atom AMP, and Sundara headphones. I feel reasonably sure that if there is any benefit from Hi-Res to be had for a consumer like myself it should be noticeable from that equipment. I've only done two single-blind studies where my wife would play the same track on Amazon HD or Qobuz (lossless and maybe Hi-Res) vs YouTube Music (lossy) and I was able to pick the lossless 10 for 10 on two separate occasions, but I'll be the first to admit that the lossy YouTube Music sounded great too.

If I get some time I'll try and repeat the test using my speaker system, but I feel less confident in my ability to tell a difference there. A) Because I feel like headphones are better for critical listening and are less affected by the dynamics of the room etc. and B) My speaker system is great for movies, but only good for music. It's a Onkyo 7.2 receiver, Pioneer Andrew Jones towers with matching center, a couple of generic in ceiling speakers for surround (which wouldn't come into play here anyway), and two 10" Polk subs. I could be wrong, but I highly doubt that system in my living room with porcelain tile flooring and a bunch of windows is going to produce super accurate sound for a test like that.

To summarize I feel like personally my headphone setup is superior to my speaker setup and also I feel like headphones in general are better for picking out details and critical listening, unless a person has a really acoustically good room with superb equipment. But that's just my opinion/experience.

If I give it a try sometime I'll post my results, lol.

What's your opinion on speakers vs headphones for critical listening?

 

[Kal Rubinson]

What's your opinion on speakers vs headphones for critical listening?

I find it hard to tolerate headphones for any listening due to their distorted spatial presentation (unless with a binaural source).
Your home theater room/system is probably not of the caliber of your headphone set-up.

 

[MRC01]

... To summarize I feel like personally my headphone setup is superior to my speaker setup and also I feel like headphones in general are better for picking out details and critical listening, unless a person has a really acoustically good room with superb equipment. But that's just my opinion/experience.
...
What's your opinion on speakers vs headphones for critical listening?

I generally agree: either can be better, depends on the equipment and the listener. Reference quality headphones are easier and cheaper (even expensive headphones are cheaper than speakers + room + preparation + amp), but they have more individual/personal variation in how people perceive the sound. A speaker system can be just as revealing as the best headphones, with less variation, more consistency in how different people perceive them, but getting there requires greater effort & expense to do it right.

 

[danadam]

1. Is it worth paying for 24 bit audio files with 48khz or higher, as opposed to CD quality tracks with 16 bit at 44khz?

A) In my humble opinion, yes, but barely. Let me explain. [...] I haven't done a test between Hi-Res and CD,

That's what I got from your answer

 

[RayDunzl]

Regarding high-res, I consider 44-16 to be musically transparent, but not perceptually transparent. By this I mean: certain sounds (jangling keys, square waves, and some others) when recorded in high res (say, 96-24) and then downsampled to 44-16, can be differentiated in DBT.

This creates a question in my mind for which I don't find a quick answer:

Is it necessary to band-limit the higher rate file before downsampling, or not?

Is there a difference in the lower rate files if you do or you don't?

(If this is an ignorant question, I apologize for my ign'ance)

 

[dc655321]

Is it necessary to band-limit the higher rate file before downsampling, or not?

Is there a difference in the lower rate files if you do or you don't?

Yes, higher rate file must be bandwidth limited to Fs/2 of the lower rate file.
Otherwise, aliasing artifacts will be present in down-sampled file.

 

[danadam]

Is it necessary to band-limit the higher rate file before downsampling, or not?

Is there a difference in the lower rate files if you do or you don't?

Yes. For example, a 42 kHz tone at -18 dBFS:

sox -r96k -n -b16 tone42k.flac synth 5 sin 42k norm -18

downsampled, without filtering:

sox tone42k.flac -r48k downsampled.flac downsample 2

produces a 6 kHz tone:

 

[MRC01]

That's correct, when converting to a lower sampling rate, bandwidth limiting is done first. That's a necessary step, so most resamplers do this automatically. You shouldn't have to filter yourself beforehand.