[KeithPhantom]

See the video link in my sig for a demonstration of what -40dB actually sounds like. For the purposes of playing music in the home, that is inaudible.

I didn't want to express my disagreement with this, but as a person at the other side of a device that allows us to communicate, I would like to express that I do not agree with you on this statement. -40 dBFS (assuming 0 dBFS to be a reasonable 80 to 90 dB SPL) is noticeable with music and test tones (watch Ethan Winer's video where he does this experiment). A DAC with distortion in this range may or may not be different than other DACs depending on the frequencies involved, the intermodulation/harmonic distortion, and the noise level. This is supported by the collective effort of ASR that measured and did listening test with an occasional ABX, where it was found that many horribly measuring DACs (SFDR around 40 dB) either didn't change the sound or just did it slightly. There are outliers in every situation (just read the paper where 3 out of 32 subjects could detect in lab-perfect conditions a signal of 28 kHz, but I consider this an outlier, not definite evidence of ultrasound detection capabilities by humans), and these are expected out of every situation, even from many established opinions in science. Science will change if evidence is presented, the methodology is scrutinized, and results are replicated, science isn't static.

Sorry @castleofargh I wanted to express my opinion, I'm not talking about ultrasonics or inaudibility, I'm now restricting myself to talk about bit depth in this specific thread.

 

[bigshot]

I think he's Canadian, maybe? And is probably just himself. If he isn't, you're still making statements/allusions about it without proof, in Sound Science.

The forum displays IP addresses to mods. That is the way to test it. I always kept an eye on IP addresses when I was modding forums back in the day. It's very useful.

Perhaps it's two knuckleheads coordinating their efforts over PM. It's the same thing. If he isn't a troll, he's the other thing. No real difference between those two things in practice.

-40 dBFS (assuming 0 dBFS to be a reasonable 80 to 90 dB SPL) is noticeable with music and test tones (watch Ethan Winer's video where he does this experiment).

Ethan Winer uses the absolute worst case scenario... a horrible buzzing sound under very quiet cello music alternating on and off. It doesn't get any worse than that. And -40 is the last point where the sound is really at all audible under the music. With pop music, it wouldn't be audible. With distortion that followed the modulations of the music, it wouldn't be audible. If it was continuous your ears would quickly hear past it.

I agree with you that under some worst case circumstances -40dB is audible. But your living room probably has a noise floor that high. If you take your portable rig out, street noise will be much higher. You deal with noise greater than that all the time and don't even think about it. People listen to LPs with noise in that range and claim it sounds better than CDs. There is such a thing as not perfect but good enough to do the job.

Audio people always want to draw hard and fast lines to define what is "acceptable" and what isn't. Then they sit down and think of unlikely extreme cases to give them an excuse to push that line a little further. Then an even less likely one to push it a little further... pretty soon we are in Amir territory. It's important to keep a handle on *relative importance* when we talk about thresholds.

-40dB is about the last point that it even matters at all any more- and then only in worst case situations. And the since decibel scale is logarithmic the difference between -40dB and -96dB is massive. No reason at all to worry about CDs. And noise and distortion levels in modern cheap amps is just as low if not more so. What's the problem here? There are much bigger fish to fry if you want great sound that this stuff. That's my point.

 

[jlawler]

sox -V3 -b 16 rate -v -L 44100
sox -V3 -b 24 rate -v -L 44100

https://gofile.io/d/XzTems

foo_abx 2.0.6d report
foobar2000 v1.5.1
2020-07-19 20:56:20

File A: 09 Don't Bring Me Down.flac
SHA1: a19b2743680e319eaee3c8bcc08867e475848723
Gain adjustment: -3.06 dB
File B: fake24 Don't Bring Me Down.flac
SHA1: ca846c3afe2c4447b1355e542f2666ba7d120c72
Gain adjustment: -3.06 dB

Output:
WASAPI (event) : DENON-AVR (2- NVIDIA High Definition Audio), 24-bit
Crossfading: NO

20:56:20 : Test started.
20:56:40 : 01/01
20:57:03 : 02/02
20:57:24 : 03/03
20:57:48 : 04/04
20:58:23 : 05/05
20:58:45 : 06/06
20:58:58 : 07/07
20:59:16 : 08/08
20:59:16 : Test finished.

----------
Total: 8/8
p-value: 0.0039 (0.39%)

-- signature --
ee60e9789ab7aa2df76c1e6f78cc6ff5b203c7f8

 

[dazzerfong]

sox -V3 -b 16 rate -v -L 44100
sox -V3 -b 24 rate -v -L 44100

https://gofile.io/d/XzTems

foo_abx 2.0.6d report
foobar2000 v1.5.1
2020-07-19 20:56:20

File A: 09 Don't Bring Me Down.flac
SHA1: a19b2743680e319eaee3c8bcc08867e475848723
Gain adjustment: -3.06 dB
File B: fake24 Don't Bring Me Down.flac
SHA1: ca846c3afe2c4447b1355e542f2666ba7d120c72
Gain adjustment: -3.06 dB

Output:
WASAPI (event) : DENON-AVR (2- NVIDIA High Definition Audio), 24-bit
Crossfading: NO

20:56:20 : Test started.
20:56:40 : 01/01
20:57:03 : 02/02
20:57:24 : 03/03
20:57:48 : 04/04
20:58:23 : 05/05
20:58:45 : 06/06
20:58:58 : 07/07
20:59:16 : 08/08
20:59:16 : Test finished.

----------
Total: 8/8
p-value: 0.0039 (0.39%)

-- signature --
ee60e9789ab7aa2df76c1e6f78cc6ff5b203c7f8

Just had a look at the files - one's in 96000 Hz, the other is 44100 Hz.

If you're testing for bit depth only, keep the sample rates the same. Otherwise, and not surprisingly, different sample rates are interpolated and sample slightly by DACs. In extreme cases, like with this random old DAC I have lying around, the music is 'sizzly' with 96000 Hz source as it downsamples to 44100 Hz poorly.

If I was cynical I'd say that your AVR says what the sampling rate is on its readout, but I'm willing to take the benefit of doubt here. If you're interested, do the following:

- Start with the same source file
- Only touch the bit rate - leave the sampling rate the same. Bump one down to 16-bit, then open that 16-bit and save it as a 24-bit file.
- Save both as WAV files. That way, the files are equal in size and there's no compression whatsoever (lossless or otherwise)

Have fun!

 

[bigshot]

I think you need someone to independently verify your results. I would like to see you do this with a test conducted by an independent person. I'm sure you can get volunteers here in this group if you let us know where you are located.

I don't use this ABX software. How would you go about gaming it? Like this? https://hydrogenaud.io/index.php?topic=119399.0

If I was cynical I'd say that your AVR says what the sampling rate is on its readout

My AVR has a handy screen that shows exactly what it is playing right there on the TV in front of you. Wouldn't even take 3 minutes to burn through a test like this that way.

 

[dazzerfong]

I think you need someone to independently verify your results. I would like to see you do this with a test conducted by an independent person. I'm sure you can get volunteers here in this group if you let us know where you are located.

I don't use this ABX software. How would you go about gaming it? Like this? https://hydrogenaud.io/index.php?topic=119399.0



My AVR has a handy screen that shows exactly what it is playing right there on the TV in front of you. Wouldn't even take 3 minutes to burn through a test like this that way.

His method makes no sense - if you change the volume, you by nature changed the file. The file hash will be completely different.

If he was inclined to cheat it, he'd have to reverse-engineer the hash algorithm of foobar's ABX plugin. Doing a file swapperoni would be obvious through the hash. Until proven otherwise, not going to accuse him of that.

 

[bigshot]

I'm Mac so I don't know how Foobar works. But I see people saying that the purpose is to find out for yourself, not proving anything to others. This guy is bound and determined to prove he can tell a difference in less than 20 seconds. I know how close that really is. The readout of the AVR then.

 

[gregorio]

[1] It is very unfortunate the CD format is only 16/44. It is very limited.
[2] Of course there is a big brain washing that 16/44 is enough.
[3] They have to sell CDs.
[4] Poor misguided children who have to listen digital format instead of analogue.

1. It is indeed very limited but that is NOT "unfortunate", it's the opposite!

2. Indeed, there was a lot of marketing that 16/44 was enough. Then there was a lot of marketing/brain washing that it wasn't enough and Hi-res was required. So, how can we tell which is brain-washing/marketing BS and which is the actual facts/truth? Hint: The answer is the name of this subforum!

3. True but they also have to sell hi-res playback equipment and content.

4. You have this backwards, even school children are taught that we listen to acoustic sound waves, not digital or analogue audio. And, didn't you know that "digital format" is converted from/to analogue?

[1] Why should I have to enter a gauntlet in order to have a conversation?
[2] I don't think it encourages dialogue or community.

1. You don't have to "enter a gauntlet" in order to have a conversation. However, if you are going to make assertions of fact that are contrary to the established science/facts, that's an entirely different matter and OF COURSE you have to "enter a gauntlet"! If you didn't have to "enter a gauntlet" then human knowledge would be based SOLELY on whatever anyone wanted to claim, which of course is why science exists in the first place. It has evolved into a strict, onerous "gauntlet" (called the "scientific method") which represents humankind's most accurate and reliable method of separating fact from fiction, that discourages even the most sophisticated charlatans.

2. Absolutely, that's the point! More precisely, it's to discourage the dialogue of false information being presented as fact and obviously, if this in turn discourages a community that relies on false information presented as fact, then that's their problem.

sox -V3 -b 16 rate -v -L 44100
sox -V3 -b 24 rate -v -L 44100
https://gofile.io/d/XzTems

Firstly, the two files you posted appear to be of slightly different length and only null by about -20dB! I don't have access to my studio and usual tools but I took your 96kHz original, converted it myself to 16bit (at home on my old laptop and freeware editor) and it nulled beyond the limit of Audacity's meters (-60dB).

Secondly, it's trivially easy to discern 16bit from 24bit! Just find a particularly quiet part of the song, usually the last one or two seconds where it fades to digital silence, whack the volume up by about 30dB or so and it's easy to discern a difference, even with internal laptop speakers. Of course though, +30dB does NOT comply with the condition of "reasonable listening levels". If you tried to listen to the whole song at +30dB either most of it would be nothing but total distortion or you'd damage your drivers or ears (as explained in the OP)!

I'm not accusing you of deliberately falsifying evidence, maybe you just made an inadvertent technical error with the files and/or inadvertently made the listening methodology error just described?

G

 

[dazzerfong]

I'm Mac so I don't know how Foobar works. But I see people saying that the purpose is to find out for yourself, not proving anything to others. This guy is bound and determined to prove he can tell a difference in less than 20 seconds. I know how close that really is. The readout of the AVR then.

Seems like you searched 'ABX cheat foobar', saw the title of a thread that suits your beliefs, and went with it.

EDIT: called it!

See here's the thing I don't get. You demand proof for anything, someone offers it, instead of scrutinising you immediately throw it out and accuse someone of cheating without any discourse. With the ivory tower you've made, what's the point in engaging in discourse?

 

[bigshot]

The reason I doubt is because you are saying that you can do in 20 seconds what no one else has been able to do under controlled conditions. If you want to follow the scientific method, the next step for you is to prove you can discern a difference when someone else is overseeing the test. If you aren’t cheating, I would suggest that you contact someone at the AES that can get your results published for peer review. You don’t inspire confidence in me, and the way to get me to accept your results is to do a double blind controlled listening test. Other people may be nice and give you the benefit of the doubt, but I won’t believe it just because someone with ten posts and a pseudonym on an Internet forum says it. But proving it to me in this group isn’t important. You should prove it in peer reviewed circles.

It was pretty easy to find that with a search engine wasn’t it. When did you first find that page?

 

[jlawler]

1. You don't have to "enter a gauntlet" in order to have a conversation. However, if you are going to make assertions of fact that are contrary to the established science/facts, that's an entirely different matter and OF COURSE you have to "enter a gauntlet"! If you didn't have to "enter a gauntlet" then human knowledge would be based SOLELY on whatever anyone wanted to claim, which of course is why science exists in the first place. It has evolved into a strict, onerous "gauntlet" (called the "scientific method") which represents humankind's most accurate and reliable method of separating fact from fiction, that discourages even the most sophisticated charlatans.

The guantlet has been baseless accusations aimed at debasing me & discrediting me as a person - that is not part of the "scientific method."

2. Absolutely, that's the point! More precisely, it's to discourage the dialogue of false information being presented as fact and obviously, if this in turn discourages a community that relies on false information presented as fact, then that's their problem.

I have not presented any false information.

Firstly, the two files you posted appear to be of slightly different length and only null by about -20dB! I don't have access to my studio and usual tools but I took your 96kHz original, converted it myself to 16bit (at home on my old laptop and freeware editor) and it nulled beyond the limit of Audacity's meters (-60dB).

I used standard tools (sox) in the manner recommended & disclosed my config parameters in case they needed any adjustment. I made no modifications beyond that. If you want to provide the 16bit version you made I can retest.

Secondly, it's trivially easy to discern 16bit from 24bit! Just find a particularly quiet part of the song, usually the last one or two seconds where it fades to digital silence, whack the volume up by about 30dB or so and it's easy to discern a difference, even with internal laptop speakers. Of course though, +30dB does NOT comply with the condition of "reasonable listening levels". If you tried to listen to the whole song at +30dB either most of it would be nothing but total distortion or you'd damage your drivers or ears (as explained in the OP)!

I'm not accusing you of deliberately falsifying evidence, maybe you just made an inadvertent technical error with the files and/or inadvertently made the listening methodology error just described?

G

I did not change the volume nor did I use any tools to measure / display the bit rate or sample rate. I just used my ears and the difference is very obvious to them.

 

[KeithPhantom]

Secondly, it's trivially easy to discern 16bit from 24bit! Just find a particularly quiet part of the song, usually the last one or two seconds where it fades to digital silence, whack the volume up by about 30dB or so and it's easy to discern a difference, even with internal laptop speakers. Of course though, +30dB does NOT comply with the condition of "reasonable listening levels". If you tried to listen to the whole song at +30dB either most of it would be nothing but total distortion or you'd damage your drivers or ears (as explained in the OP)!

So this is the reason I can hear the noise floor of many recordings, I know it's this because all songs in an album have a particular hiss, and if I start listening to another one, the hiss is different. It has being annoying to say the least. I listen at around 80 to 90 dB SPL with headphones. I've checked ground loops and faulty equipment, but even with my phone used as a source/DAC/AMP I can do it. I'm maxing out the digital volumes and using my amplifier as volume control. It also happens with music from Spotify and in multiple OS. I have transcoded my library twice (FLAC-ALAC-FLAC+downsampling and dither for files >44.1/16), it could be this.

 

[bigshot]

So this is the reason I can hear the noise floor of many recordings, I know it's this because all songs in an album have a particular hiss, and if I start listening to another one, the hiss is different.

That is most likely tape hiss or air conditioner noise in the recording venue. Gregorio is talking about boosting the volume +30dB over normal listening level. That means boosting the volume to the point that it is uncomfortably loud. People have been known to gain ride their tests and focus on the ring outs of the song to goose the volume enough to hear a difference. Foobar is not intended to be proof to anyone except the person taking the test. If someone wants to cheat, they can cheat. But ultimately, they're just lying to themselves. If you have to cheat to prove your point, why not just switch to a point that is defensible?

I've done hundreds of 24 vs 16 listening tests. I do one every time I finish a mix to make sure the master file is the same as the mix I am approving. This is not generally an issue that is questioned except by audiophiles who either have never taken the time to test for themselves, or are so deeply invested in their argument they feel the need to cheat to make their point.

If someone comes along and has superpowers and says "it's easy to tell the difference", I'm not going to just automatically believe them because they say it. People say all kinds of things that aren't true on the internet all the time. If their results are that much of an anomaly, I'd suggest that they get their results independently verified. They can feel free to get all huffy and offended and say "How dare you question my honor!" I really couldn't give a flying flip about that. We're all just avatars and typing. The internet isn't the field of honor. If you want respect, you have to earn it... and that takes more than just 10 or 15 posts trolling an internet forum.

If it's that easy to perform an extraordinary feat, an independent person with experience in the field should be conducting the test and observing the results. If it's true, it's a lot more important than just an argument in an internet forum. It should be shared with the scientific community and peer reviewed. Of course, if it's a lie, it's not worth anyone's time at all...

I never offer my own tests as proof to anyone else but me. My goal is to improve my own sound system and find out for myself. I'm not a research scientist. I encourage everyone to do controlled tests *for themselves*. It doesn't matter what other people think. People who are too invested in proving things to other people raise my suspicions to be honest. It smells like someone who is out to prove their point at all costs. If he gets mad and huffy, maybe he'll go away. He isn't interested in listening to anything anyone else says, so that would be the best of all options.

However, if this guy can actually hear 24 bit as distinctly different than 16, it doesn't change the huge body of research on this topic that has already been done. All it means is that for him there is a difference. That doesn't mean that it makes a lick of difference for anyone else in the world. If he wants to prove 24 is necessary, he needs to prove that lots of other people can tell the difference too.

And he needs independent verification.

 

[bigshot]

All of this is silly anyway, because he's claiming to hear greater than -96dB on a recording made in 1976 on 24 track analogue tape that didn't have a noise floor that low. It's kind of obvious what's going on here...

Brrrrrruuuuuuucccceeeee...

 

[jlawler]

What is silly is that some who "claim" to adhere to science are really more akin to religious zealots when presented with evidence that may call into question their beliefs.

It is not your right to smear me & as long as the mods allow it I will keep defending myself. Actually, I will just ignore you from now on as you obviously have nothing positive to offer.