I remember reading on 3D soundsurge how a problem with consumer DACs is that they are only designed to reproduce voltage up to whatever corresponds to 0xFFFF when they need to be able to reproduce sound to a higher voltage than that to reproduce the recording accurately.

Is there any software that will let me process a wav file to reduce its volume level so that this doesn't happen?

I believe you are referring to an audio phenomenon known as 'clipping'. If a recording has already been clipped, there really isn't much that can be done.

No, there is another sort of clipping that may occur when DAC reconstructs original wave signal from descrete data.
Detailed explanation of the problem can be read in the following paper:
http://www.cadenzarecording.com/papers/Digitaldistortion.pdf
(thanks to krabapple for the link).

Foobar2000 with replaygain can eliminate any clipping, it also can be used as a hard limiter.

actually, foobar's hard limit is hard-coded to limit to fullscale. meaning it's still quite possible for a DAC to clip on it's output.

i wouldn't be worried though. any clipping produced in the DAC will be but a needle in a gigantic stack of needles. also, clipping of the duration of 2/44100 samples or less is not going to be audible unless you've got super-ears (and then i doubt it).

i believe the BBC did some testing in the '50s about how long a clip you can get away with, and they concluded that clips could actually be quite long before anyone noticed them (5ms?) so long as the clipping is occasional and not consistent. a clipped sine would be very audibly different from a smooth one, but a clipped transient could be quite transparent. i don't know how detailed this research was though. it may be poppycock

In a related question, if I am putting together my own CD from various wav files, is there any way I can roughly equalize the volume level among the wav files before burning them?

http://en.wikipedia.org/wiki/Replay_Gain
http://wiki.hydrogenaudio.org/index.php?title=Replaygain

Digital Distortion
That's probably why Waves UltraMaximizer L1/L2 (for example) have a template for final CD mastering, limiting samples to -0.3dB from FS (-0.2dB in L2) thus preventing consumers' DAC overload during interpolation between two "hot" samples.

ReplayGain is of no use for the application I have specified. Where have I implied that I am using foobar2000 for CD playback? In fact, I never have and never will.

Use Wavegain.

but replaygain IS of use. use foobar2000's integrated cd burning tool, installed with the special version. Burn with replaygain, and you should be ok, considering you can limit the volume to 89db, a 3db range (IIRC) which should be plenty.

Wavegain would still be the easiest solution. Use it on the WAV files before burning. It's permanent, and afterwards you can use any software you like to burn the CD. There's no dependency on any other package.

It would be very difficult for dnewhous to use foobar to burn with Replaygain if he's not using foobar. Using something and not using it at the same time is a very complex feat. I've attempted it for years, but with no success.

now now.. he said that he did not use foobar for CD playback, nor that he ever would. That doesnt mean he can't use it to play other things, or to burn with it

Even though it does exactly what you asked for?



foobar2000 has nothing to do with Replaygain, aside from the fact that its one of many applications that impliments the standard.

I was about posting a question about DAC interpolation clipping, when - after searching - i found a thread...
I found here an empiric answer : Post #3
The paper linked talks about (even more than) a 6 dB margin to avoid that phenomenon on extreme samples. Have anybody widely studied this before, or made global figures ?

I'm dealing with a huge work I think, but "why not" ? (kidding)

I was setting foobar main volume to -0.20/-0.40 dB to avoid this*...

Thanks in advance

*PS : By the way, could someone be able to tell me the difference between "that" & "this" (I'm french)... please

Cheers,
Christophe

[Out of subject] PS2 : That (this?) link could have served a recent "hot" thread about normalizing...

I'd love to hear more about this as well. Just thinking about it, I'd guess that most oversampling DACs would have this problem, but I've never tested it.

Regardless, this is essentially a less extreme version of the typical clipping problem incurred with lossy compression and rounding errors in the decoder. Using replaygain should completely eliminate the problem.




'This' refers to things that are close by or recent (in space, time, topic, etc). 'That' refers to things which are distant:

"This chair (that I'm sitting on)"
"That chair (thats across the room)"

in one of the manuals, i think for either L2 or L3, they clearly say the signal can be clipped by the cd player by as much as 5 db. so even a ceiling of .3 may sometimes not be enough.
ps - it should be noted that most quality cd players should in fact output the correct level even if the disc is normalized to full scale with no headroom at all. however really cheap ass cd players may not do that, hence the additional distortion. but for me it will be a problem only if i can actually hear the distortion.

I just started encoding lossless, with WavPack; so i ran a bit away (and, maybe in this case, in an easier way) from my old MPC's and ReplayGain expectations... Thanks 2Bdecided by the way.

I think it would be quite stupid to implement oversampling without predicting the clipping introduced, since it all comes to the same "naive" seeing! I hope it has "pushed" companies to anticipate this default (especially this one could go up to 6 dB over full canonical digital range and maybe over). Listening to Muse albums whose drums sound like "low bitrate encoded MP3's", i wonder if they did?! It's a pity since their music is...

Wouldn't it temporarily (audibly ) represent/destroy more energy than one could have "preserved" by dithering or intermediary bit depth increase - in case of "pre-"(it has to be said) and even post-mastering?

Sorry, I'm going to sleep, i'm getting too low on batery... My cat will confirm this

THANKS Mike Giacomelli for your course (In french I've got "ici"==this & "là-bas"==that ; I hope I will manage... )

Good night ! (quite unusual, but...) cheers !
Christophe (little drunk... i don't remind in fact. )

Hmm, what about "digital" amplifiers, shouldn't they be free from this clipping problem?