[vorbis] APPLAUD.WAV problems

Michael Smith msmith at labyrinth.net.au
Sun Feb 17 15:11:12 PST 2002

>> > The reason applaud.wav is a problem (if it is at all anyway) is that
>> > there is an upper and/or lower bound bitrate setting for quality modes
>> > for vbr encoding
>> This is not true, the encoder does no such thing.
>Yes it is, you should run some actual tests and see for yourself. 

Gian-Carlo was stating a fact, not an opinion. Both he and I have
read the code - we _KNOW_ what it does, so we can state that the encoder
does not do that - nor, with rc3, does it have the capability to (rc4
will be able to, though it won't by default).

>> Moreover, the problem is that the encoder is not aware that it is 
>> creating artifacts, so it will not use a higher bitrate because it 
>> does not realize something is wrong, i.e. it's psymodel is not
>> perfect (this also pretty much invalidates your entire mail)
>> -- 
>> GCP
>before you go throwing the entire mail away I would suggest encoding
>absolute silence.  It's fun, especially when you get about the same
>bitrate as the cd audio clips you're encoding at the same quality
>level.  The evidence seems to be pointing at a lower bound being set by
>the quality level and a detailed look at every single block shows that
>there's a mysterious lack of bitrates below around the 128kbit range in
>ogg files encoded at -q 5  Now you might be quick to jump and say, well
>you need at least that to keep the stereo lossless and I really have no
>evidence to counter that. But then I'd be quick to show you that -q 4.99
>exibits the same exact behavior as only the lower bound bitrate is a bit
>lower, around the 112kbit range.  

Absolute silence encodes (always) at around 600 bps (_bits_ per second.
i.e. under 1 kbps). 
For sections with complex content (not neccesarily very complex, but
not silent, and not absolutely trivial), the bitrate will typically
be fairly close to the average - that's just how the encoder works.
I wonder how you inspected the bitrate per block? None of the tools
or players I've seen will tell you that. (well, tools I and other
developers have written to test various things, but not widespread
tools that non-developers would have).

Whilst some of your arguments have some limited value, your conclusions
are simply completely wrong.


<p>--- >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.

More information about the Vorbis mailing list