Best LAME options for encoding 48khz Options

[dewdude]

I recently got the DVD that came with The White Stripes - White Blood Cells...and the audio on the entire DVD is 48khz PCM..I've ripped the two audio only tracks to wav..but I'm wondering..what's the best command line option for encoding 48khz content..I DO NOT want to convert to 44.1 so don't preach to me about it...my sound card and my MP3CD Player play 48khz just fine.

[Delirium]

Well, frankly, I'd recommend either converting to 44.1 kHz or using something other than LAME to encode it. The only high-quality LAME modes are the --presets (--alt-presets) and they're not tuned for 48 kHz. The quality loss in 48->44.1 dithering will be far less than the quality loss you'll incur by using untuned lossy compression on the 48 kHz audio.

[Jebus]

He said don't preach about converting to 44.1kHz.

Failing that... (And i do suggest you do it) just try a good 'ol --alt-preset standard and see what happens... It shouldn't resample unless you tell it to. No idea if it'll sound good though. Or do a custom "-vbr -V2 -h --lowpass 19" perhaps.

[magic75]

Isn't the most important thing to get the best possible quality? From what I have understood from other threads downsampling to 44,1 kHz and encoding with lame -aps should give better results (or at least safer results). And if you lowpass by using -aps or --lowpass 19, what is the point with using 48kHz anyway?

[dewdude]

I'm a believer in the more samples per second the more pure your resulting wave will be. Let's face it....44.1 is probably more than I can hear anyway...but...if I've got a 48khz source..I'd rather keep it 48 khz. If it's not tune for 48 khz, then fine..I'll keep it WAV if I have to. I could always go WMA9Pro.

[2Bdecided]

I think you're talking about samples per second, rather than bits per sample. Or maybe not?

48kHz gives more samples per second than 44.1kHz. For uncompressed audio, you have a fixed number of bits per sample (e.g., 16, or 24, or even 8 - yuk).

For compressed audio, people also sometimes talk about the number of bits per sample. Coding a stereo 44.1kHz audio signal at 128kbps gives 1.45 bits per sample. Coding a stereo 48kHz audio signal 128kbps gives 1.33 bits per sample.

So, at 48kHz, using a constant bitrate codec, you've actually got fewer (coded) bits available to represent each sample (or group of samples) from the original audio signal than you would have at 44.1kHz. A variable bitrate codec could increase the bitrate to accomodate for this - but alt preset standard hasn't been tuned or even checked to make this happen.

However, if it was me, I would try it and see what happens. It might just be OK. The thing is, resampling to 44.1kHz, and then encoding will be OK. Your choice.

Cheers,
David.

[tigre]

I had a similar problem a while back (DVD 48kHz PCM track -> mp3). Doing a forum search all threads I found claimined the same: Encoding at 44.1kHz with lame --alt-presets gives better results, even if you upsample to 48 khz again on playback. I found two samples in the track where I could successfully ABX 48kHz lame 3.90.2 aps vs. Original but not 44.1kHz lame 3.90.2 aps (downsampled with SSRC before encoding, 24bit output file). Unfortunately I don't have the original anymore, so I can't provide samples. The difference was some slight ringing/chirping added. In spectral view some "dropouts" (black "holes") at ~ 15kHz were visible at these positions (I did this after ABXing ).

Are you sure that your MP3CD player doesn't resample or resamples in a decent way if you feed it with non-44.1kHz mp3s? To be sure there's no aliasing I would test this first by encoding some high pitch sine tones/sweeps to mp3 at different sampling rates and play it back.

I would encode the whole pcm losslessly (e.g. using Monkey's Audio) for archiving/pc playback and encode it to mp3 in the way that seems most confidence-building to you. If you notice later that there's something wrong with the mp3 you can still encode it again (maybe lame 3.94 is available then, optimized for 48kHz - just dreaming ) using the lossless file as source.

This post has been edited by tigre: May 6 2003, 13:25

[KikeG]

Yeah, in addition to what 2Bdecided says, lossy encoding will result in much more degradation than a good (SSRC) resampling to 44.1 KHz.