[Ryokan]

The OP of the 24bit vs 16bit thread might help you.

G

Thank you I'll have a look later, it looks an interesting read.

 

[WoodyLuvr]

The OP of the 24bit vs 16bit thread might help you.

G

Very nicely presented/written post. A very nice brush up on the subject. Thank you for the link.

 

[WoodyLuvr]

Results are out... very interesting data and thoughtful conclusions. Well done and well presented @Archimago !

RESULTS: Internet Blind Test of 24-Bit vs. Dithered 16-Bit Part Deux - Daft Punk Edition

 

[bigshot]

Those Slovak nations are really good at cheating on tests!

 

[ostewart]

I couldn't hear a difference, and great read in terms of the results.

 

[xiled10]

Late to the party and gave it a try. I don't have a super fancy system but I could not tell the difference

 

[gregorio]

Results are out... very interesting data and thoughtful conclusions. Well done and well presented @Archimago !

TBH, it’s just a re-hash of numerous other such tests done over the last 25 years. Meyer & Moran did the same with SACD vs 16/44, published by the AES in 2007, same result even without any noise-shaped dither and they also used test samples identified by audiophiles as “the most obviously different”.

I did exhaustive testing in the late ‘90’s, as did many/most engineers and there’s two problems the “believers” would have to overcome: Firstly, there is no rational explanation for how there could be any audible difference and Secondly, in controlled testing (where participants couldn’t cheat), no one has ever been able to hear a difference (at reasonable listening levels).

25 years or so and still not a shred of reliable evidence or a rational explanation for why the difference could be audible, so why is this long settled question still being asked? Answer: There’s money to be made, therefore there’s marketing and a vocal percentage of people suckered by it!

G

 

[71 dB]

There’s two problems the “believers” would have to overcome: Firstly, there is no rational explanation for how there could be any audible difference and Secondly, in controlled testing (where participants couldn’t cheat), no one has ever been able to hear a difference (at reasonable listening levels).

Considering that the difference of 24 bit and 16 bit (made from the 24 bit version using dither as one should in digital audio) is dither at a level so low it isn't audible in any real life listening scenario by itself, how could anyone hear it when it is masked by the main signal? Even if dither wasn't used when converting the 24 bit version into a 16 bit version and we would have granulating quantization noise as the difference signal, it would still be at a level too low to be heard in any real life listening scenario.

the noise floor (dither) of 8 bit audio is easy to spot, but each additional bit makes noticing the dither significantly harder and in my opinion at 10-11 bits the dither starts to be rather hard to notice, but this depends on the nature main signal. Dynamic classical music makes it easier while DR 6 pop music would probably be fine with only 8 bits. That's why I have been saying 13 bits is enough in consumer audio. It takes the dither low enough to not be audible in real life listening scenarios. Now, 16 bit digital audio has 18 dB lower dither level than 13 bits! Anyone with two functioning brain cells should realise such dither can't be audible meaning 16 bit and 24 bit must sound identical as long as placebo effect doesn't do what it does.

25 years or so and still not a shred of reliable evidence or a rational explanation for why the difference could be audible, so why is this long settled question still being asked? Answer: There’s money to be made, therefore there’s marketing and a vocal percentage of people suckered by it!

G

This issue seems to be linked to ego for many: People convince themselves they can hear differences so small hi-res audio is needed and marketing + placebo effect gives them wrong confirmation bias. They genuinely thing we 16 bit-listeners can't hear as well as they do and that our listening gear is crappy.

 

[gregorio]

Considering that the difference of 24 bit and 16 bit (made from the 24 bit version using dither as one should in digital audio) is dither at a level so low it isn't audible in any real life listening scenario by itself, how could anyone hear it when it is masked by the main signal?

Two connected answers to that point:

1. The dither isn’t always masked by the main signal. Sometimes there is no “main signal”, for example pauses/silences in the music or the main signal is too low in level to mask the dither. For example, with fade-ins/outs or the end of reverb tails.

2. What exactly do you mean by “any real life listening scenario”? A fairly common “real life listening scenario” when I’m working is a looped very quiet section with the amplification whacked up. In which case, during those examples in point #1, the 16bit dither noise can be quite easily audible. Of course, this real life listening scenario is an ear damaging level for consumers, who are just listening to songs/tracks and not looping only the short, quietest sections with the amp whacked-up.

G

 

[71 dB]

Two connected answers to that point:

1. The dither isn’t always masked by the main signal. Sometimes there is no “main signal”, for example pauses/silences in the music or the main signal is too low in level to mask the dither. For example, with fade-ins/outs or the end of reverb tails.

2. What exactly do you mean by “any real life listening scenario”? A fairly common “real life listening scenario” when I’m working is a looped very quiet section with the amplification whacked up. In which case, during those examples in point #1, the 16bit dither noise can be quite easily audible. Of course, this real life listening scenario is an ear damaging level for consumers, who are just listening to songs/tracks and not looping only the short, quietest sections with the amp whacked-up.

G

Excuse me if I sense a bit reluctance to understand what I mean. However, I'll explain myself so nobody should remain confused about what I wrote.

1. It is self-evident that masking doesn't happen when there is nothing to mask anything! However, even in those cases we might have a 200 seconds long track for example of which say 10 seconds are pauses/silences/etc. meaning 95 % of the time masking happens.

2. Well, I certainly did NOT mean you working on a track in studio! I have said it 100 times here that for consumer audio 16 bits is enough (even 13 bits! for that matter), but in studios in music production more bit depth is beneficial. I have taken this tread as purely concerning consumer audio and that's why what I said does apply only to music consumers. I would never say you should limit yourself to 16 bits* while working in a studio! That would be rather silly in 2023! Even 24 bit dither is quite audible if you "whack it up" by 100 dB! However in typical consumer audio listening scenario you are not supposed to "whack it up" hardly at all!

Hope that clears things up.

* Having more dynamic range than 16 dB equivalent allows "better" 16 bit dithered truncations, because the dynamic range "fades out" into the noise floor softly similar to analog audio.

 

[redrol]

I think I don't trust this test. Either show the exact resampler or whatever, or Im going to assume its a trick.
[edit] I have audio editors. Zoomed into the files, saw no extra bits.

 

[redrol]

 

[bigshot]

It’s a test of bit rate, not sampling rate.

 

[redrol]

I see what you mean. Im not sure how to find out if this is real or not.

 

[bigshot]

Do research to understand how digital audio works. In my sig is a link called CD Sound Is All You Need. I've pointed to it many times in the past. Everything you need to know about high sampling and bit rate audio is there.