Ripping old CDs to MP3 using LAME - Resample? Options

[BFG]

Quick question to those who know a LOT more about high-quality lossy rips than I do.
Is there any perceivable benefit (or, perhaps, disadvantage) to upsampling CDs--which typically are encoded at 44.1khz - to 48khz?

I've compared a handful of tracks at both sample rates and, so far, can't tell any difference between the two.
Right now I'm running LAME 3.99.5 -v -V0 -q0 --replaygain-accurate.

[saratoga]

Is there any perceivable benefit (or, perhaps, disadvantage) to upsampling CDs--which typically are encoded at 44.1khz - to 48khz?

No, and in fact its a pretty bad idea unless you need 48k output for some reason, and even then I would recommend resampling after decoding, not before encoding.

[BFG]

No, and in fact its a pretty bad idea unless you need 48k output for some reason, and even then I would recommend resampling after decoding, not before encoding.

Thanks for the quick reply. Is it a bad idea because many players won't play 48k correctly, because it'll diminish the quality/transparency on the players that can, or something else?

[AndyH-ha]

Upsampling adds nothing but an increased storage space requirement. Since the normal reason for lossy encoding is to decrease bandwidth and storage requirements, upsampling could be said to be a "bad" idea or at least a not very bright idea.

As far as effecting audio quality, it is probably a total null, but it is possible that the codec works better for 44.1kHz than for 48kHz. I don't recall reading anything about that, it probably isn't true, but having no interest in it, it is the sort of thing that might go in one ear and out the other for me.

[BFG]

Upsampling adds nothing but an increased storage space requirement. Since the normal reason for lossy encoding is to decrease bandwidth and storage requirements, upsampling could be said to be a "bad" idea or at least a not very bright idea.

As far as effecting audio quality, it is probably a total null, but it is possible that the codec works better for 44.1kHz than for 48kHz. I don't recall reading anything about that, it probably isn't true, but having no interest in it, it is the sort of thing that might go in one ear and out the other for me.

Hmm...that does make sense. Since the original recording was at 44.1khz, you're really only converting that to 48khz, or 96khz, or whatever, rather than actually creating a 48khz etc. recording out of what you started with.
I suppose it's a bit like creating a 320kbps MP3 from a 192kbps MP3--you don't gain anything, and just use more storage space.

I do wonder if LAME would better handle preecho issues (and other rare problems) at a 48khz sample rate while encoding than at 44.1khz, but considering it's mostly keyed to 44.1khz I'll probably leave it set as is.

Thanks for the info.

[greynol]

What's with the accurate replay gain scanning? Are you using one of the tiny fraction of apps/plugins that actually makes use of RG info in the Lame header?

This post has been edited by greynol: Oct 17 2012, 04:49

[halb27]

...I do wonder if LAME would better handle preecho issues (and other rare problems) at a 48khz sample rate ...

Yes, in principle pre-echo issues should improve a bit. However tonal issues get worse a bit. With this in mind I wouldn't allow for an additional audio process at the highest VBR quality level.
If you care about highest possible VBR quality you might consider using -V0+ or similar with my 3.99.5z variant.

[BFG]

What's with the accurate replay gain scanning? Are you using one of the tiny fraction of apps/plugins that actually makes use of RG info in the Lame header?

It's more for my own interest than anything. I like to compare the LAME calculation to what I get through MP3Gain. And on my machine it only adds a half second or so of processing time.

[BFG]

If you care about highest possible VBR quality you might consider using -V0+ or similar with my 3.99.5z variant.

Thanks for the tip. The changes you implemented for -V0+ (and -V0+eco) do seem interesting. I'll have to give it a spin.