[vorbis-dev] vorbis-utils features ;)

[Beni Cherniavsky]

On Thu, 22 Mar 2001, Gregory Maxwell wrote:

> On Thu, Mar 22, 2001 at 01:47:45AM -0500, Keith Wright wrote:
>  
> > In that case it is "easy" to write a filter for the input that
> > squelches at whatever level is appropriate to your application,
> > and it seems obvious to me that that is the Right Way to do it.
> 
> This is called noise gating and it's really too simplistic to be that much
> of a benifit.
> 
> Currently vorbis does a very smart, but consertive thing:
> It doesn't know how loud you will be playing a file, so it can't assume it
> can just throw out or quantize the hell out of quiet stuff. However, it does
> know that the loudest sound within a sample ill not be louder then a certian
> amount so we do have an upper bound on the volume level you will have it set
> to. However, on quiet bits (like analog silence)  this is too consertive
> (because there is no loud reference), I suspect that future encoder will be
> smart enough to assume the listner won't be acting as a constant human
> compressor and changeing the volume knobe between every block, thus it could
> assume that the knob is set to no more then foo-dB louder then it was at
> loudest point prior in the song. Then the rest of the algorithim can take
> care of representing the unaudible bits with according quality.
> 
This looks a Good Idea - add an option to the encoder to assume a "limited
knob reaction speed".  The way I understand it now, it takes the maximum
value of the current block as the reference - the option would say to take
the maximum value over the last T msecs...  Lookahead is probably
meaningless (OK, his ears are going to hurt badly 15 msecs from now but he
doesn't know that yet and will be upset at Vorbis' low dynamic range which
was the last thing he heard before becoming deaf :-).

All this is assuming one won't do strong band-filtering to listen closely
to quiet frequencies of the sound but Vorbis assumes this already.  
People should be punished for certain quality-problem-hunting listening
habits ;-)


-- 
Beni Cherniavsky <cben at crosswinds.net>
                 (also scben at t2,cben at tx in Technion)

10001110111100111100001001010 m/s

--- >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-dev-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.