> From: [EMAIL PROTECTED]
> Date: Mon, 3 Jan 2000 14:45:49 +0100
>
> Hi Everyone,
>
> all the best for the new year etc...
>
> Perhaps this is an idea to improve the Short/Long block switching:
>
> When would you probably want to switch from long to short blocks? Probably when 'the
>amount of pure sinusoids' is low... in other words, when the signal is kind of noisy.
>Wouldn't it make sense to use a spectral flatness measure; not necessarily the one
> in the MPEG standard for a detection of switching? An idea would be: filter the
>signal/mask ratio, determine the difference between this ratio and the unfiltered one
>and switch whenever the difference is larger than ...!
>
> Erik
> --
Hi Erik,
Just in case you haven't seen this, a great test case for
short block switching is testsignal2.wav (under the quality section
on the LAME web page). The MPEG/ISO algorithm fails quite badly on this
test case, and most of the work for the LAME short block switching
algorithm was debugged against this case.
Mark
-=- MIME -=-
Hi Everyone,
all the best for the new year etc...
Perhaps this is an idea to improve the Short/Long block switching:
When would you probably want to switch from long to short blocks? Probably when 'the
amount of pure sinusoids' is low... in other words, when the signal is kind of noisy.
Wouldn't it make sense to use a spectral flatness measure; not necessarily the one
in the MPEG standard for a detection of switching? An idea would be: filter the
signal/mask ratio, determine the difference between this ratio and the unfiltered one
and switch whenever the difference is larger than ...!
Erik
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
