P2P-Zone - View Single Post

Snarkridden · 07-06-02, 02:02 PM

The minimum setting really depends on the material being encoded, it is true to say that LAME adjusts the bit rate to that which required, so in theory you could set 32k as the minimum, though with most music, I doubt that the 32k minimum would ever be reached, except at the fade in/fade out points of the track, so you are unlikely to actually see very much actually encoded at that level, however a lot of P-2-P applications seem to use the first few frames to determine the actual bit rate to report, which of course would be wildly wrong on material with a un-demanding start.

So setting a practical lower rate just below the actual minimum required, does help with this aspect of file detail reporting.

If your material is really hi quality, i.e tracks from a SACD disc ot a hi-sample DAT recording (96k) then using VBR will throw away too much of the fine detail in the High frequency regions, so best to use a fixed bit rate anyway.

If you can do a spectral analysis of your WAV file from a CD, then encode it VBR, decode it to a WAV file again, do a spectral check on this second WAV file, you will easily see the limitations of VBR
on high quality material.

Do the same tests with LAME 3.92 at fixed bit rate 320k, and you will find the decoded WAV file will have almost an identical spectra to the original probably up to about 21.5 kHz a perfect match with the original.

I am still playing with this SACD player at the moment, material to test with is hard to come by, most being slightly tarted up original CD material, so true tests will have to wait for a while.

By the way, NEVER use Joint Stereo on high quality material, the quality reduction, is just to much for the small drop in file size.

Snark..

Snark..

07-06-02, 02:02 PM	#3
Snarkridden OpenNap Server Operator Join Date: Jan 2002 Location: U.K Posts: 401	Why Indeed? The minimum setting really depends on the material being encoded, it is true to say that LAME adjusts the bit rate to that which required, so in theory you could set 32k as the minimum, though with most music, I doubt that the 32k minimum would ever be reached, except at the fade in/fade out points of the track, so you are unlikely to actually see very much actually encoded at that level, however a lot of P-2-P applications seem to use the first few frames to determine the actual bit rate to report, which of course would be wildly wrong on material with a un-demanding start. So setting a practical lower rate just below the actual minimum required, does help with this aspect of file detail reporting. If your material is really hi quality, i.e tracks from a SACD disc ot a hi-sample DAT recording (96k) then using VBR will throw away too much of the fine detail in the High frequency regions, so best to use a fixed bit rate anyway. If you can do a spectral analysis of your WAV file from a CD, then encode it VBR, decode it to a WAV file again, do a spectral check on this second WAV file, you will easily see the limitations of VBR on high quality material. Do the same tests with LAME 3.92 at fixed bit rate 320k, and you will find the decoded WAV file will have almost an identical spectra to the original probably up to about 21.5 kHz a perfect match with the original. I am still playing with this SACD player at the moment, material to test with is hard to come by, most being slightly tarted up original CD material, so true tests will have to wait for a while. By the way, NEVER use Joint Stereo on high quality material, the quality reduction, is just to much for the small drop in file size. Snark.. Snark..