Hello Robert,
Sunday, May 21, 2000, 9:05:19 PM, you wrote:
RH> Hi Roel,
RH> the speed drop in Version 3.83 is because of the noise calculation
RH> in scalefactor band 21 (at 44.1 kHz samplerate it represents the
RH> frequency range from 16 to 22.05 kHz). If there is some noise in that
RH> scalefactor band and it is greater than the noise our ATH for that
RH> band allows, that band is distorted.
RH> In Version 3.83 there were some useless tries to get rid of that
RH> distortion, but we can't get rid of that distortion, cos we don't have
RH> a scalefactor to color it. Tonight I checked in some code to do some
RH> earlier stopping in that situation, so the speed should be at
RH> the old Version 3.70 level again.
thanks ;).
btw: I hope you didn't undo what _was_ accomplished by 3.83:
- sweep got fixed
- http://users.belgacom.net/gc247244/extra/takehiro-1.png
Is one example (consistent with other findings, check the "Bush" graph on my
http://users.belgacom.net/gc247244/analysis.htm) that 3.83 fixed
_alot_ of weird results in 3.70 (AND 3.80, but same graph, so no
update)
--
Best regards,
Roel mailto:[EMAIL PROTECTED]
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
