Is there any music that actually requires 24 bits for replay?

[Human Bass]

Like @sergeauckland says, no point whatsoever.

The best of the best DACs we've seen so far can resolve roughly 20 bits. I don't think most people have any idea of how crazy big a span 120dB dynamic range gives you.

You'd need a listening room with a ridiculously low noise floor + a set of speakers that can play brutally loud without compression to get there, and it won't be pleasant.

Yes, the loudness wars went way too far, but there's a reason why it started. A truly huge dynamic range will have you contantly reaching for the volume control, every time it switches between unintelligible soft and painfully loud. More realistic, but also really annoying.

I bet most of the audiophiles, who waffle on about dynamic range, have never heard the real thing. Their prime examples of such a playback situation probably doesn't equate to much more than 10 bit when you do the actual math.

Last of Us 2 game has audio option for huge dynamic range and can get really annoying indeed lol.

 

[Multicore]

Is it not the case that 24 bit has advantages when using DSP?
Sure I read it somewhere, and would be most grateful if someone could remind me why?
Thanks in advance!

The implementation of various algorithms and equipment can be easier with more DR. I have a looper that uses 32-bit float and that makes complete sense given that levels can be very hard to predict. Some ADCs and recorders now have 32F. But the finished product for distribution to consumers always fits in 16.

 

[Chr1]

Yes, but what if the consumer uses DSP?

Pretty common here. And doubtless will become more widespread surely.
At least for those seeking optimum sound quality.

 

[Matt_Holland]

A lot of this is for marketing purposes in the same way that 4K video resolution is pointless on a 50” TV and 8K is silly, but makes people go, ‘oooooh’.

The audiophile community who eschew science still believe that the “staircase” is audible, so to them, the more bits and samples the closer we get to analogue. This ignorance has been exploited by hifi manufacturers for decades now and continues to this day.

 

[BeeKay]

A lot of this is for marketing purposes in the same way that 4K video resolution is pointless on a 50” TV and 8K is silly, but makes people go, ‘oooooh’.

The audiophile community who eschew science still believe that the “staircase” is audible, so to them, the more bits and samples the closer we get to analogue. This ignorance has been exploited by hifi manufacturers for decades now and continues to this day.

 

[Multicore]

Yes, but what if the consumer uses DSP?

Pretty common here. And doubtless will become more widespread surely.
At least for those seeking optimum sound quality.

Then the algorithm should convert to 24 or whatever, appropriately scaled for what it's going to do, and then convert to whatever is needed for delivery to the next stage. We consumers should not have to be concerned about this, just like I feel no need to check on all the other aspects of design inside the box: tolerance of resistors, brand of capacitors etc.

I use the word "should" because in an incompetent design, a.k.a. a badly engineered product, it might not be the case. My WiiM has a handy EQ feature but don't set any of the sliders to a positive number or you risk clipping. Most of the material it plays is normalized close to full scale and the output is optical. To my mind it shows incompetence in algorithms and/or UI to implement it that way.

 

[Chr1]

So are you saying that there's no advantage to using 24 bit with DSP?

Apologies, I am not clear on this. I thought there was.

 

[manisandher]

So are you saying that there's no advantage to using 24 bit with DSP?

Apologies, I am not clear on this. I thought there was.

Not if the DSP is well-implemented. For example, Roon will convert 16-bit to 64-bit, perform any processing, and then convert back to 16-bit.

 

[Chr1]

Thanks. I guess now the question is: how do you know if your DSP is well implemented?

Personally I am using MathAudio RoomEQ via Foobar2000. May have to look into this.

 

[Multicore]

So are you saying that there's no advantage to using 24 bit with DSP?

There can be advantages to the use of more bits internally in the implementation of the audio device that uses DSP. But even when the device is using 24 bits internally, there's no practical difference between using 16 or 24 bit formatting for the input signal to that device (or for its output).

Apologies, I am not clear on this. I thought there was.

Perhaps this distinction between internals and interfaces is the cause of your not being clear.

As Amir's chart at #13 shows, there's not much of cost to using 24 bit digital interfaces and that's part of why there's a lot of them around, so you can put hi-res or hi-def marketing on it and so create the need for upgrades. But everything should work equally well with 16-bit recordings. In other words, you don't need to waste money, bandwidth or storage space on the extra bits per sample when obtaining audio files to play. You can if you want. You can buy fancy speaker cables if you want and they should work just fine, with the same meaning of "should" as above, i.e. assuming correct designs.

 

[Multicore]

Thanks. I guess now the question is: how do you know if your DSP is well implemented?

I generally don't.

Personally I am using MathAudio RoomEQ via Foobar2000. May have to look into this.

If you like. But the caveat I and @manisandher gave about assuming correct engineering applies to everything, not specifically these extra bits. The same caveat applies to all the engineered products we buy. While I sometimes find bugs, I do not generally task myself with design verification of consumer products. And if you're not expert in the tech then you may not accomplish much, although examining designs can be useful if you're trying to learn.

 

[manisandher]

As another very quick example, you can take a 16-bit file, attenuate by 90dB (in something like Audacity, which uses 32-bit processing), amplify again by 90dB, and the resulting file will null perfectly (better than -300dB in DeltaWave) with the original.

 

[Owl]

I went to my local used record store yesterday and most of the shoppers were teens, and twenty something, buying records not CD's. Whether there after the cool album covers or, "chasing sound" I don't know.

 

[Chr1]

Chasing fashion more like.
As teenagers do.

 

[kemmler3D]

This thread is full of good explanations... If there is any difference between a 16 and 24 bit version of a track, the differences only start to appear below -96dBfs... If you've ever heard anything that quiet in a recording, your system was probably turned up to a scary level.

 

[Frank Dernie]

This website derides dynamic compression, but in fact the marketplace rewards it.

Two things make it good: old ears, and listening at low levels.

Recorded music is not reality; it is like a photograph, painting, or sculpture. A thing unto itself.

I spent thousands of hours in Photoshop, and what I’ve learned is that dynamically squeezed photographs “pop” and are more likely to be considered good or exciting.

This is a playback issue. We do not want highlight or shadow detail eliminated. We want it made visible in the presentation medium.

The flip side of this is, all other things being equal, we prefer media having wide dynamic range. Light emitting monitors are more vivid than projection screens; both are more punchy than paper.

I do not agree with your analogy with photography.

I think compression on music recordings ia about making it listenable on modest equipment like little smart speakers and in noisy environments, earbuds on a 'plane or a busy family home and not wanting to annoy other family members.

I have loads of my original master tapes with no compression and whilst performances are amateur the sound is more realistic than most record releases IME.

Turning the volume up makes the sound more dynamic so selling records with a high average level sounds good on comparison with full dynamics unless the average volume is matched (ie almost never) and efficient speakers sound more dynamic until you match levels which IME is never done right by eye and volume control.

I can't say what is preferred in a photo, manipulated a lot may well be it

 

[Ropeburn]

For me personally, I use digital volume control, therefore I like to use 24-bit transmission path to give myself comfort that the original 16-bits are more or less preserved, doesn’t matter if the lowest bits are audible or not.

And for 24bit content, I comfort myself that bits 25++ are not audible

Digital volume control is done in 32bit these days. The source bitdepth has no consequence.

 

[Frank Dernie]

This thread is full of good explanations... If there is any difference between a 16 and 24 bit version of a track, the differences only start to appear below -96dBfs... If you've ever heard anything that quiet in a recording, your system was probably turned up to a scary level.

Absolutely.
In my hairy arsed experiments to estimate how good (low distortion and noise) sound needed to be for me to hear it I started off by playing music at my normal listening level then turning down the volume. I could still hear -something- at -60dB but barely and I came to the conclusion, maybe erroneously, that were something at that sound level be playing when I was listening to music at my normal level I wouldn't hear it - thereby satisfying myself that for me less than 0.1% is good enough.

Also lets accept that plenty of people on other sites believe LPs are superior SQ to digital and even the most cloth eared would not be able to believe that if more than 11-bits or so was actually needed for music.

 

[boxerfan88]

Digital volume control is done in 32bit these days. The source bitdepth has no consequence.

Yup, I know. 64F for foobar 64bit.

The 24bit “transmission path” - I was referring to the output side, when 64F is converted to 24bit for transmission to my DAC.

 

[Chr1]

I will confess to being slightly OCD about getting 24bit files a few years back. Took me a while to realise that the quality of the actual recording/mastering is way more important than the bit depth of the playback file.