Rendering MP3's from 32 bit audio

[pdarg]

I need some clarification from someone who is an expert in digital audio/rendering/etc.

I have a 32 bit .WAV file master. I need to encode it to MP3. I can easily do this in WaveLab - but the question is: do I need to dither and render to 16 bits first? Am I missing an important step by rendering a 32 bit .WAV file (without dither/bit conversion) straight to MP3?

Thanks in advance!

[pdarg]

No one? Really?

[Elteto]

Does WaveLab even support direct rendering to mp3 from a 32-bit .wav? Does it ask you to dither the source file first, or does it do it automatically? If it automatically dithers first, are there any settings you can control the dithering and noise-shaping process with?

Also, does WaveLab have its own mp3 codec, or does it rely on an external codec library? (I would suggest using the Lame library for mp3 encoding, its psycho-acustic model yields better results than other mp3 encoders. It is also free.)

Reading around on some related sites, technically, Lame supports rendering an mp3 from a 32-bit .wav source, but there are some stipulations.

Please read this related discussion:

http://www.hydrogenaudio.org/forums/index.php?showtopic=63360

It is a bit of a read, but is a great eye opener, and it specifically mentions the dithering output of WaveLab, comparing the spectrograph outputs of various converters.

[Exalted Wombat]

I need some clarification from someone who is an expert in digital audio/rendering/etc.

I have a 32 bit .WAV file master. I need to encode it to MP3. I can easily do this in WaveLab - but the question is: do I need to dither and render to 16 bits first? Am I missing an important step by rendering a 32 bit .WAV file (without dither/bit conversion) straight to MP3?

Thanks in advance!

I think Wavelab does this quite transparantly? I'd just let it get on with the job. You're degrading the material to MP3 for goodness sake! Dithering has a subtle effect, quite swamped by the greater damage you're aiming to do.

[pdarg]

Thank you both for your responses.

I am using WaveLab 4 - which has its own MP3 encoder built-in. I use the Lame encoder at 256 kps and 44.1kHz.

WaveLab 4 is able to encode a 32 bit file to MP3 without any prompting for dither - and no problems. To me, it seems like the way to go, however, I have received conflicting advice on this. I suspected that no dither was needed; I believe that you have confirmed my belief here.

[Martin Walker]

WaveLab 4 is able to encode a 32 bit file to MP3 without any prompting for dither - and no problems. To me, it seems like the way to go, however, I have received conflicting advice on this. I suspected that no dither was needed; I believe that you have confirmed my belief here.

I'd tend to agree - Wavelab is very good at conversion, and an MP3 converter doesn't specifically specify a dithered 16-bit input signal.

Martin

[ramthelinefeed]

if it really concerns you,
do it both ways (dithering first, and not bothering to dither),
then invert the phase of one of the resulting files, line them up to start at the same place in your DAW, and play them both at once. You will then hear the difference, whatever it may be.

Whether you wish to max a 12" remix out of the resulting difference is up to you.

[AllyB]

Audacity can handle this just fine and it uses the Lame encoder for Mp3's too.