Importing 16-bit wave and exporting to constant rate 320 kbps MP3. No processing is being done. Just import then export.
Does this create a bit depth change that would introduce rounding errors that dither would be helpful for?
Thanks
There’s no need to dither. MP3 doesn’t have “bit depth”. But of course, it is lossy compression so it does alter the data.
…Normally you can’t hear dither (or the effects of dither) anyway, so as a practical matter it doesn’t make any difference one way or the other. At 8-bits, it probably makes a difference.
I normally leave dither enabled (“shaped”) all of the time. When exporting to MP3, dither will have a negligible effect whether enabled or disabled.
Strictly speaking, dither is unnecessary in the case that you describe, so you can turn it off if you want to.
Personally, I’m unconvinced that it is ever worth dithering when exporting to MP3. The encoder (LAME) should be able to handle any supported bit format without “pre-processing”. Applying dither does make the audio content more “complex” and for VBR encoding it tends to make the file size a little bit larger.
Thanks for the great explanations. I’m in the camp of not using things (dither in this case) if they don’t have a meaningful benefit. I’ll go dither-less on this.
Thanks.
Perhaps I should explain that comment a bit. In MP3s, dither does still do what it is designed for; namely it prevents harmonic distortion caused by quantization, but if you are really bothered about getting the very best sound quality, you shouldn’t be using a lossy compressed format. For a format (MP3) that only produces an approximation of the audio, the question of whether to dither or not is hardly relevant.
Since no Audacity processing that alters sample values is being done. Would there be any point/benefit of dithering regardless of whether exporting to lossy or lossless?
Thanks
Assuming that the original audio was 16-bit and, as you say, no processing that alters sample values, then dithering has no benefit.