[vorbis] A killer clip

[Aleksandar Dovnikovic]

Hello Monty,

Saturday, 28 July, 2001, 07:38:34, you wrote:

M> Oooh, *good* one.  Best killer clip anyone's found yet.
M> (peering at it through psytune now...)

Here's what MP+ author said on this issue:

***********

> This is familiar problem to me. This problem is caused because
> of clipping. 

exactly! when lowering the amplitude by about 5-6 dB the distortion
vanishes. as there are no internal clippings it may also be possible
to lower the amplitude while _decoding_ (the float-values
themselves do not contain these distortions, the clipping is added
within the 16bit truncation). maybe it would be possible to detect
such situations while encoding through an analysis/synthesis-loop
(will cause higher cpu-load).

...

the proceeding would be to encode a frame (choosing the desired
quantizers and quantize the samples) and decode it again
(re-quantize to float-values and synthesize the frame). if the
decoded samples would show clippings the resolution of the most
critical bands must be increased. so, there would be several
loops necessary if there are clippings in the decoded samples....
of course the bitrate for such signals will rise :)

***********


-- 
Vorbis Xtreme | http://solair.eunet.yu/~aldov/
Ogg Vorbis is the free, open source alternative to MP3

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.