24-bit and 16-bit audio comparison – audio test results

What quality of sound can a person determine by ear? In a very recent audio test, respondents are asked to blindly distinguish between sounds with a 24-bit and 16-bit dynamic range. Each of them downloaded several pairs of 24-bit files, one of which underwent a 24-16-24-bit conversion, that is, in practice it was a 16-bit file. They were asked to determine the difference.

The test involved 140 volunteers (138 men and 2 women – an honest demographic for audiophiles). Average age of respondents: 44 years.

According to the questionnaires, more than 20% of the respondents called themselves musicians and sound engineers, so it is possible to compare the results between “professionals” and amateurs, taking into account the statistical error.

The cost of audio equipment for survey participants typically ranges from $ 1,000 to $ 3,000.

The survey results for the three pairs of files are quite curious. In two of the three compositions, the correct and incorrect answers were distributed exactly in half.

And in the composition of Bozza, 52.85% of users made a mistake by mistaking a 16-bit file for a 24-bit one.

20 respondents answered all questions correctly and 21 people made a mistake in all variants, which also fits the statistical distribution.

It’s even more surprising that musicians perform worse than average, even taking into account statistical error! Vivaldi’s composition was especially confusing.

And here is the result among users whose audio equipment costs expensive: They could not distinguish.
Headphones also don’t help distinguish 16-bit from 24-bit music.

Summarizing. Of course, there are applications where it is necessary to work with 24-bit audio (the same mastering). But the fact is that 16-bit and 24-bit audios are completely indistinguishable from each other by ear. If someone claims to be able to hear the difference, that person is probably wrong.

16-bit vs. 32-bit audio

Is it useful to make recordings at 32 bits?

No, and nobody can. It’s fun to think that everyone can. Bit depth is all about margin, that’s it. So if you have 24 bit depth you have about 140 dB of headroom. Who Needs It Are we logging jet engines from neutral to finish? Maybe there is someone. But most home musicians and ALL studios that make music never use that kind of margin. Most music has life compressed by someone while being mastered, and only uses the beats on top of the output. So why do people use 24-bit registers? Is it possible to make large files? It does not help the quality of the recording in any way.

The sample rate gives you more frequency response. And that’s it. However, at 44.1 kHz, the records are good down to a whopping 22 kHz, and that’s beyond ANY analog gear of the past. People claim that if you have the ability to turn up the frequency response, then the harmonics go into the audible and “affect the sound.” As far as I know, this has never turned up on double blind hearing tests to make a difference. Again, for the average home studio or these days, music (pop / rock metal tends to be pretty good) is created so badly that any … even possible frequency response gain is destroyed. And furthermore, very few people on Earth can hear beyond 20 kHz and THERE IS NO MUSIC there.

Anything greater than 16 bit 44.1 kHz is a loss of data because the data is simply not used. I think the idea of ​​”more is better” is behind the idea that “higher bit rates and higher sample rates” sound better because they are bigger. It’s bullshit. This is digital data. There are no slipping and slipping. Either there is enough bandwidth to receive the data and play it back (and there is with 16 bit 44.1 kHz) or there is none. Fortunately, if people insist on using higher speeds, no one will die. They are just wasting data space.

Answer 2:
No.

32-bit: choose your taste

Actually, there are two 32-bit types that are used in music. The one you find in a DAC or ADC is basically the same as your 24-bit DAC, with only 8-bit resolution. However, it is primarily “because we can, not because we should.” So I just looked at a 32-bit random DAC, “PCM1795 32-bit 192kHz sampler, extended segment, stereo D / A converter” from Texas Instruments. Scroll down to page 7 of the specification, in the Electrical Specifications section, and you will get the signal to noise ratio. Regardless of the sample rate, it is 123 dB for stereo, 126 dB for mono. That is actually 20.5-21 bits of resolution. So there is absolutely no difference between this DAC in 24-bit and 32-bit mode. In 24-bit mode, you have 20.5-21 bits of true audio, 3.0-3.5 bits of “marketing”. In 32-bit mode, you have 20.5-21 bits of true audio, 11-11.5 bits of marketing.

I’m not even sure why anyone was concerned about 32-bit DACs. When I first went into ADC / DAC with 16-bit stereo audio, it was amazing. Ultimately, my computers will have sound that is “as good as you ever need it”. Well not really, and even then it was pretty clear that you could create an audio system with SNR better than 96dB. I rarely needed it, because before CD no sound reproduction medium offered this level of noise, but it was possible. But nobody makes a system that generates 144 dB in the analog world.

There is another type of 32-bit audio sample that is used in music, this is a 32-bit numeric field, which includes a sign bit, a 23-bit mantissa, and an 8-bit exponent. It’s just a 24-bit audio sample with an added exponent field. Exponential means that the audio can be processed multiple times with virtually no loss of resolution. But if you listen to it, it turns back to a whole number. You cannot hear 32-bit floating point audio; it must be converted to integer format for playback.

But even a 24-bit sample is more than necessary. This is what is now the recording standard. If you could play a 24-bit audio file on an amplified system tuned so that a sample value of 0x0000 0000 0001 only matches the human hearing threshold, the full scale value would exceed the pain threshold and could cause real damage . your audience. This is a digital dynamic range of 144 dB.

I say digital because you can’t think of it as analog. In fact, a really good amp can give you a range of 120dB or so. You won’t be able to fully reproduce your signal