• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Can two software players sound different from each other and why?

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,609
Location
Central Fl
Who cares about "bit perfect playback" when it means you can't use Replaygain? It only matters if you can't trust your software stack to not ruin your audio streams.
Do you really want to do Replaygain to your 16/44 lossless or better files?
 

q3cpma

Major Contributor
Joined
May 22, 2019
Messages
3,060
Likes
4,416
Location
France
Well, of course. Volume normalization and resolution or even sound quality have nothing to do with each other.

Technically, you could apply the gain and reencode to get your bitperfect playback, but this gives another problem.
 
Last edited:

Eirikur

Senior Member
Joined
Jul 9, 2019
Messages
318
Likes
509
Do you really want to do Replaygain to your 16/44 lossless or better files?
Why not? As long as it is processed 24bit or better and sent out as >=24bit to the DAC.

Precise application then gives an average amplitude quantization error per sample of only 1/(2^23) (half wave) = 1.19e-7.
Applied to a 4 volt differential signal that is a maximum error of about 0.00048mV - I can live with that (barring calculation errors).

At the higher frequencies (>11kHz) there is of course a compound reconstruction error because of the sparse sampling, but look at the margins! This might actually be a valid application of upsampling, just to keep the compound gain or attenuation error ridiculously low!
 

Eirikur

Senior Member
Joined
Jul 9, 2019
Messages
318
Likes
509
Well, of course. Volume normalization and resolution or even sound quality have nothing to do with each other.
I don't think that is strictly true, you will always incur the amplitude quantization error. If done in the 44.1/16bit domain it will certainly have some (measurable) impact. Just imagine the case where all samples in a certain section are rounded up, you'd get a (small) DC bias!
 

Eirikur

Senior Member
Joined
Jul 9, 2019
Messages
318
Likes
509
Well, yes, but that implies no dithering.
Agreed, although dithering will "kill" the bottom ~3dB, so again not "bitperfect".

In any case, no-one in his/her right mind would use the constrained 16bit domain in a modern software environment making it all this a bit academic.
Still, it's interesting to do the calculations to get a feel for the absolute error margins. In Audacity I managed to lose almost 6dB to noise when also correcting for DC bias! All of this starting with a "perfectly" generated 20kHz 44.1/16 tone at 80% volume.

Bottom line: keep your source as is, use a proper player with enough bit resolution, and normalize to your hearts content on playback.
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,609
Location
Central Fl
Each to there own :)
 
Top Bottom