MP3 conversion 320cbr to v0/v2 Options

[lostguru]

pretty much the title...

if I have 320 cbr MP3 files (I'm going to assume all from lossless sources), would converting directly from 320 cbr to v0 or v2 result in lower quality than if I have converted directly from lossless to v0 or v2?

This post has been edited by lostguru: Aug 15 2012, 17:11

[goa pride]

i converted all mp3 dj set from 320 to vbr V0.5
Do you recommend V0 with Y switch enable istead of V0.5?

[Dynamic]

i converted all mp3 dj set from 320 to vbr V0.5
Do you recommend V0 with Y switch enable istead of V0.5?

Are you now talking about
1) converting from lossless source (original CD or FLAC etc) direct to MP3 or
2) from one MP3 setting (320 CBR) to another (-V 0.5 or -V 0 -Y)?

lostguru has been talking about (1), which is the best approach, the rest of this thread had been about (2) which is known as lossy-to-lossy transcoding.

Your answer will help us to advise you.

If it's (2), I'd suggest that -V 0 -Y would be better if those are the only two options as it provides a fraction more headroom above the level normally required for transparency in the frequency range where it's most likely transcoding issues will become audible. I really doubt that sfb21 (the frequency range above 16 kHz will reveal transcoding artifacts).

If it's (1), I'd imagine both settings would be essentially indistinguishable, already having a good deal of safety margin beyond the threshold of transparency, which is probably nearer to -V3 or -V2, and again, in real music, whether -Y is used or not, is unlikely to make an audible difference.

For case (2), -V3 or -V2 might be a little risky, because the inaccuracy of two encoding steps is being combined, and they're aimed at pretty much being just enough to ensure transparency (problem samples aside), with little margin of safety when encoding from the original, lossless source. Having said that, for a number of uses I've happily done just that sort of thing and not noticed a problem, but if you're DJing, you don't want to be taking chances.

[goa pride]

i converted all mp3 dj set from 320 to vbr V0.5
Do you recommend V0 with Y switch enable istead of V0.5?

Are you now talking about
1) converting from lossless source (original CD or FLAC etc) direct to MP3 or
2) from one MP3 setting (320 CBR) to another (-V 0.5 or -V 0 -Y)?

lostguru has been talking about (1), which is the best approach, the rest of this thread had been about (2) which is known as lossy-to-lossy transcoding.

Your answer will help us to advise you.

If it's (2), I'd suggest that -V 0 -Y would be better if those are the only two options as it provides a fraction more headroom above the level normally required for transparency in the frequency range where it's most likely transcoding issues will become audible. I really doubt that sfb21 (the frequency range above 16 kHz will reveal transcoding artifacts).

If it's (1), I'd imagine both settings would be essentially indistinguishable, already having a good deal of safety margin beyond the threshold of transparency, which is probably nearer to -V3 or -V2, and again, in real music, whether -Y is used or not, is unlikely to make an audible difference.

For case (2), -V3 or -V2 might be a little risky, because the inaccuracy of two encoding steps is being combined, and they're aimed at pretty much being just enough to ensure transparency (problem samples aside), with little margin of safety when encoding from the original, lossless source. Having said that, for a number of uses I've happily done just that sort of thing and not noticed a problem, but if you're DJing, you don't want to be taking chances.

2) for dj sets, available for free on Soundcloud, none use VBR V0, most are 320k, none is using lame 3.99.5
1) for any tracks found on Soundcloud that are flac
i use XRECODE with external Lame 3.99.5 x64

This post has been edited by goa pride: Sep 9 2012, 21:32

[Dynamic]

2) for dj sets, available for free on Soundcloud, none use VBR V0, most are 320k, none is using lame 3.99.5
1) for any tracks found on Soundcloud that are flac
i use XRECODE with external Lame 3.99.5 x64

For 2) the Soundcloud MP3 files, why re-encode. It can only make quality worse (or no better), and if you're using -V0 or similar settings, it won't be an awful lot lower in bitrate, so why bother? The only reason to transcode is if you 'process' the files through some DSP or whatever, and want to bake that effect into the audio, or you want to do some beat-matching or mash-ups and make a new MP3 out of the result.

For 1), really any setting between -V3 and -V0 should be transparent and most folks here are happy with -V2 or -V2 -Y. Even -V5 is really very good for general music listening with most differences from the original being of the 'perceptible but not annoying' variety to most listeners in blind testing, even for most of the known 'codec killer' problem samples and with a pretty big saving on bitrate. I've also had joy at around 130kbpps with Helix VBR - the same settings that tied with LAME -V5 or -V5.7 in the 128 kbps MP3 Listening Test a few years ago - though I use that only on devices that can't play gapless. I'd rather use LAME and retain gapless compatibility otherwise.