Besides any exceptions that may arise...does LAME produce a higher quality VBR MP3 from a 48000 Hz Wav file than 44100 Hz Wav file if the LAME settings are kept the same for both conversions, or does the 48000Hz wav file have more information than an MP3 can handle?? Also, is there a way to change EAC to extract CD files to 48000Hz Wav before converting them to VBR MP3 instead of 44100Hz??

It would have to upsample the file. So I guess you're saying will a 44.1khz wav -> 48khz -> mp3 sound better than 44.1khz -> mp3? Something I can't answer cause I still don't see how people say an upsampled wav can sound "better" than the original wav itself.

NO, im not upsampling the Wav...the Wav files are coming from a DVD...they are 24 bit, 48000Hz, PCM Wav Files

Lame can take 24 bit depth input from stdin with no problems. I don't think it poses a problem from a waveform audio file (pcm) either. However, according to the nyquist theorem, the extra sampling rate of a 48000 Hz wave file is superfluous (since the normal ear can only hear up to 20 000 Hz anyways..)
In fact, 44100 Hz waves are probably in a large margin above what we need also.



Hey, who cares, right?

(suggested : resample to 44khz)

LAME is not tuned for 48kHz input. That doesn't mean it will produce garbage, just that there might be unforseen problems. Many people who rip movies encode the soundtrack to stereo mp3 with 48kHz and nobody is complaining.

Sound on Red Book CDs is stored at 44.1kHz. A perfectly upsampled signal will have the same information as the original. In reality you will have a little less.

Iirc back in the r3mix days, there were some people who tried if resampling to 48kHz could improve smearing, being that mp3 uses the same frame length at both sampling frequencies. Can anybody shed some light on this issue?

Resampling in the decoding stage can be usefull, if your soundcard internally resamples with low quality. Then it is better to do the resampling yourself with a high quality software resampler such as the SSRC resampler in Foobar.

In general, here I haven't seen any recommendation to resample the way you suggested.

AFAIK, a (normal) CD has a sample rate of 44.1khz.

If you are meaning the 48kHz files from the DVD, EAC can't extract them. If you want to upsample CD audio you will get no benefits from doing this (even if EAC could, which it can't).

Why is that some people say that upsampling makes some music sound better? As far as I know, upsampling has to do with guessing what those samples would be, if they were in a file, and depending on the algorithm, it may make the resultant upsampled file "smoother". But is a smoother file necessarily better?

You can use DVD2AVI to downsample the DVD audio part to 44 kHz. It's a pretty buggy program but it did the job for me.

If you files were originally 48kHz, I'd suggest to encode them at 48kHz instead of resampling to 44.1kHz.

I'd say, it might depend on what you intend to play the MP3's on. If they're going to be played on a platform where everything is automatically resampled to 44kHz (or 48kHz), then you might want to use Naoki Shibata's SSRC or the SSRC built into foobar2000, to produce that sampling rate, because almost certainly these resampling methods produce better quality output.

i second when NeoRenegade told.

but i would rather resample to the output neded, at playback time.

Therefor keep the original 48khz and encode it.
resampel to 44.1 khz only if you need the output and do it on playback.


There is a plugin for winamp that does the opposite,
(resamples to 48khz at playback)
doing it in software has better precision then doing it on the sound card itself. but again requires more cpu power.


most mainstream soundcards today (creative live and above) resamples to 48kKhz.... and doing a rather bad job at it compare to SSRC



to sum it up:
Keep it at 48khz if that is the original source.
Resample by software at playback... if needed


Gennerally always try to keep it as close to the original as possible, as long as possible

My thoughts are, if you're going to use it on a portable player i'd resample to 44.1kHz. Have SSRC do it, and have it output a 24-bit wav file with no dither (ssrc.exe --rate 44100 --bits 24) then feed that to LAME. This will save some space, while minimizing any issues from the resample.

If you're listening on a PC, then space isn't as much of an issue, and your soundcard is probably going to upsample back to 48 anyhow. So keep it at 48 just like Gabriel says.