johnnyV


Well I'd be interested to try the conversion software you are using. Last time I used FLAC a 30 MB waves became 26- 28 MB. That's a long ways off from 50%.
I'll agree with you second sentence 100%

bitflipper

I also believe that a 320 MP3 is more than good enough for 90% of listeners

90% is too conservative. In a blind ABX, I'd be surprised if 1% could reliably identify which file was the 320kb/s MP3 and which was the 44.1/16 wave. When I did this test myself, not only could I not hear any difference, I could not objectively measure any difference either.

I know, it seems counterintuitive. I was reluctant to believe it myself. How is it even possible to remove 88% of the data and not tell the difference? Turns out we don't hear as much as we'd like to think.

jcatena

Can you give some info on what type of stereo to choose when making an MP3. I see options for Joint-Stereo, Stereo, Variable Bit Rate etc ... What settings do you use? I use Sony Sound Forge 9.0 and set it to High Quality, 320kbs, Stereo.
 

You know what mono and stereo is. Joint-stereo consists in encoding as mono (the sum of the 2 stereo channels), and only what's different from a certain threshold is encoded as stereo. For low bit rates it usually improves perceived quality a bit while the stereo imaging is lost where it is not very obvious. If you want to avoid noticeable perceived quality losses, use stereo and 320kbps. 256 and even 196 kbps are also good. HQ VBR (variable bit rate) will result in the minimum bit rate for same perceived quality as 320kbps, depending on the material it will result in averages from 190 to 250 kbps.
Other compression algorithms provide better quality/compression ratios, the best one is AAC and is currently very widely used, since it was adopted by Apple iTunes. Most modern devices support AAC.
If what you want is lossless compression (bit accurate reconstruction after decoding), FLAC or APE are the best choices. FLAC has been some time around, APE is newer and is becoming very popular and supported by many devices, also used by iTunes.
 
> If you want to take for example a 96/24 master wav and make an MP3, is it ok to not dither and leave it as is or should it be dithered down to 44.1/16 first?

Most mp3 encoders only support 16 bit and 44k1 or 48k. But some applications will convert internally if necessary when you export an mp3. Anyway keep in mind that a sample rate conversion may add aliasing depending on the algorithm (a good one will not, but you want to be sure of that). Dithering with noise shaping should not be used prior to any processing including mp3 encoding. Dithering without noise shaping may be used, but it will not make a meaningful difference in the compressed output.