OK, Why Can't I Hear A Difference?

[oceantracks]

The simple answer is my hearing is going of course, but really...

When I switched from 24 to 16 bit dithering on the Sony Limiter (on the 2 buss) I can't hear a bit of difference. Now most of the time I'm doing pop tracks that are pretty busy...is this only going to really make an audible difference on something like a classical guitar solo or something?

Thanks
TH

[24-96 Mastering]

Originally Posted by oceantracks

The simple answer is my hearing is going of course, but really...

When I switched from 24 to 16 bit dithering on the Sony Limiter (on the 2 buss) I can't hear a bit of difference. Now most of the time I'm doing pop tracks that are pretty busy...is this only going to really make an audible difference on something like a classical guitar solo or something?

Thanks
TH

Opinions will vary on this...

For one thing, it makes more of a difference on more audiophile recordings, rather than busy pop material. Plus, how audible the bit depth reduction is will also depend on your room & monitoring setup, of course. In my room, reduction is nearly always audible, but that doesn't mean it's necessarily drastic. Most commonly on reasonably busy material, it feels like "1 very thin onion layer of soft&wide" is peeled off. (Apologies for this very esoteric sounding testimony.)

[oceantracks]

Originally Posted by 24-96 Mastering

Opinions will vary on this...

For one thing, it makes more of a difference on more audiophile recordings, rather than busy pop material. Plus, how audible the bit depth reduction is will also depend on your room & monitoring setup, of course. In my room, reduction is nearly always audible, but that doesn't mean it's necessarily drastic. Most commonly on reasonably busy material, it feels like "1 very thin onion layer of soft&wide" is peeled off. (Apologies for this very esoteric sounding testimony.)

Thanks Robin.

Yes my listening room is awful, which is probably not helping. I'll pay attention more when I'm doing a track that's very open and quiet and see what I hear...

Tom

[Ethan Winer]

Originally Posted by oceantracks

Yes my listening room is awful, which is probably not helping.

Now that's an understatement.

An untreated or poorly treated room has multiple "early" reflections arriving at the mix position soon after the direct sound from the speakers. This creates a timing mess that clouds everything, and also skews the frequency response badly. Badly is an understatement too. The graph below shows the comb filtered response you get from reflections off an untreated sidewall.

To address your original question, try the freeware +decimate VST plug-in available in the freesound bundle here:

Soundhack | freeware

This plug-in lets you reduce the bit depth to tell how low you can go before you can start to hear the degradation. For most program material you'll need to go to less than 16 bits before the bit-depth reduction is audibly damaging.

[Washington]

Well, the very idea of dithering when converting from 24 to 16 bit is to make the wordlength conversion as inaudible as possible... most particularly on low parts of the signal, like the end of a reverb tail. It's not very noticeable on a busy mix, and to A/B the dithered and undithered portion of a fading signal requires a dangerous (particularly for your monitors) amount of volume.

In other words, not hearing a difference is (basically) good news.

[space2012]

dither is shaped noise,

if your converters are plugged to dirty AC, and the powersupply does not have a verry strong AC noise reduction built in,
the AC noise will affect the noise floor in the DACs.
also the noise floor of the amplifiers of the speakers &/or headphones
get clean power.

cables, DAC design, etc...
you must have less noise & more dynamic range than the noise you are trying to hear.
16-Bits = (Bits+6.0206) + 1.761 = 98,0906dB
Dynamic range - Wikipedia, the free encyclopedia
"you need headroom."
Headroom - Wikipedia, the free encyclopedia

at least -100dB of SNR in the signal path.
Signal-to-noise ratio - Wikipedia, the free encyclopedia

if you fix all variables and still cannot hear it...
its your ear/brain.
Absolute threshold of hearing - Wikipedia, the free encyclopedia
Hearing range - Wikipedia, the free encyclopedia
Detection theory - Wikipedia, the free encyclopedia
Minimum audibility curve - Wikipedia, the free encyclopedia
Johnson–Nyquist noise - Wikipedia, the free encyclopedia

in other words...
If you have a loud PC/ventilation fan near, you wont hear dither, unless you have 5000w concert speaker system in your room all the way up.

[24-96 Mastering]

Originally Posted by space2012

you must have less noise & more dynamic range than the noise you are trying to hear.

You don't need to actually hear dither noise to notice a dithered bit depth reduction...

And I don't think you necessarily need an SNR of 98,0906dB (!) to identify the difference. A single SNR number, imo, really isn't of much use here.

As a general guess, decent acoustics & low absolute room noise, precise monitoring & a clean signal path, in that order, will be primarily where it's at (assuming your hearing is OK and you know roughly what you're listening out for). Cables & power conditioning will likely make a much smaller impact (the extent of which depends very much on specific circumstance).

[Bob Olhsson]

Dither is a choice between truncation that sounds like noise and truncation that sounds like distortion. If the distortion is inaudible, the dither is going to be even more inaudible. There is no more of an argument for leaving dither off than there is for leaving an analog tape machine's bias off.

In many monitoring situations there is no truncation which can lead people to conclude that dither is just adding noise that is supposed to bring out low level detail or something. Truncation distortion is actually that harsh cold digital sound that almost everybody hates and wants to "warm up."

Unfortunately in practice there has also been software dither that was very much broken and audible which has only added to the confusion.