[paranoia-dev] Real-time reading with paranoia ?

Monty xiphmont at xiph.org
Thu Oct 21 15:15:17 PDT 1999



> But, if I would use the paranoia library, I would somehow need to
> guarantee that I get a real-time stream.

Paranoia cannot do this.  The library is not designed for real-time behavior,
period.  Extending it to do so would require a great deal of work.

However, I submit that perhaps what you want does not require a realtime 
paranoia.  I'll submit for evidence that XMMS on my slower systems can't play 
MPEG back realtime anyway, and it doesn't bother rebuffering when it starts 
skipping.  It just buffers at the beginning.

Straight paranoia wouldn't be any worse on a slow drive.  You could also add to
the configuration a 'disable error correction' for slow systems.  The
prebuffering period could even be used as a gague of how fast (approximately)
the extraction process will be.

> It would be extremely useful if I could give the library some kind
> of deadline for the audio data, so that it can use more or less
> paranoia depending on how much time is left for the data to be delivered.
> 
> So the questions are:
> - Is it feasible to use the paranoia lib to read a real-time audio stream ?

Yes, full error correction better than real time given a fast enough drive.
However, paranoia takes no steps to guarantee realtime behavior.

> - Can I 'tune' the library to use less or more paranoia depending on how
>   much time there is left ?

Right now, you have a choice of all or nothing. Sorry.

> - How much CPU impact can I expect when paranoia is reading a real-time
>   audio stream ? (with different error-correction modes)

The error correction alg is automatic and converges to an optimum approach for
each drive.  On all but the worst drives, CPU impact should be minimal.  On bad
drives/bad disks, cdparanoia can be very CPU intensive.

Monty

--- >8 ----
List archives:  http://www.xiph.org/archives/
Paranoia homepage: http://www.xiph.org/paranoia/
To unsubscribe from this list, send mail to 'paranoia-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.



More information about the Paranoia-dev mailing list