Vorbis - Oct 2004 - Question on Quality factor, Bitrate and decode table

An HTML attachment was scrubbed...
URL:
http://lists.xiph.org/pipermail/vorbis/attachments/20041014/25f999a2/attachment.htm
-------------- next part --------------
 ?
dear All,
   Iam working on OGG VORBIS TREMOR code to port and optimize on ARM processor.
Can any one make me clear regarding the relation ship between this
 
  1. What is the relation ship between Quality factor and Bitrates?
     As vorbis is a VBR algorithm, many bitrates will be selected   
     encode and decode a stream? If so how quality factor is related  
     to bitstream? 
  
  2. In many web pages(I dont have the links to pages) related to Ogg 
     Vorbis information I read higher the bitrate we select, the 
     audio qualtiy is poor. And for the low bitrates the audio 
     quality is poor. As we go higher the bitrate, I think the 
     quality what we will get is high. But why not this in Vorbis.And 
     for the low bitrates why the audio quality is poor.  

  3. On what basis the number of decode tables are fixed for an 
     particular stream. Why the number of decode tables are not same  
     for all streams. Is any other parameter will control the number 
     of decode tables to present?

 Could anyone please give some directions/remark on this issue? Thanks very
much!


Regards,
Viswanath Deepak G

(Hm, your mail client does not interpret RFC2822 properly; the newline
characters are incorrect.)
>   1. What is the relation ship between Quality factor and Bitrates?
The Vorbis specification does not have any concept of bitrate. The
decoder does not know and does not care what the actual bitrate f an
incoming stream is.
>      As vorbis is a VBR algorithm, many bitrates will be selected   
>      encode and decode a stream? If so how quality factor is related  
>      to bitstream? 
There is no bitrate at all.  An encoder may choose to construct a
stream so that it obeys some bitrate requirements, but this is done
entirely by manipulating encoding techniques, not by making any
fundamental changes or adding explicit flags to the stream.

In general, hiher quality settings produce larger streams.
>   2. In many web pages(I dont have the links to pages) related to Ogg 
>      Vorbis information I read higher the bitrate we select, the 
>      audio qualtiy is poor.
Your English is clumsy here, so I'm not sure what you mean...

I think what you mean is that the current Vorbis encoders generally
allow you to force a stream to a given bitrate to allow uses where
bitrate controls are requires (like limited bandwidth streaming).
This results in lower audio quality, for a given bitrate, than
allowing VBR. The fundamental advantage of VBR is higher quality for
the number of bits used, the disadvantage is a lack of control over
bitrate; overall a VBR stream will be lower bitrate, but can be
subject to large spikes in bitrate when encoding complex passages.
When explicitly forcing a bitrate limit, the bitrate spikes are also
eliminated... but the complex passages will be lower quality because
they are not allowed to use more bits.  Similarly, simple passages
that would not normally use many bits in VBR will use more than are
necessary when VBR is not allowed.
>   3. On what basis the number of decode tables are fixed for an
>      particular stream. Why the number of decode tables are not same
>      for all streams. Is any other parameter will control the number
>      of decode tables to present
The tables are entirely at the encoder's whim and reflect the most
current research into codec improvements.  The tables used by an
encoder generally improve in new releases.

That said: We do support one useful codebook feature in our encoders.
For a given release, we guarantee that the tables used by the encoder
will be constant for a given set of encode settings.

Monty