Re: Digital Talking Book Standard

[Nicolas Pitre <nico@cam.org>]

On Mon, 19 Nov 2001, Peter Toneby wrote:

> On Mon, Nov 19, 2001 at 11:35:59AM -0500, Nicolas Pitre wrote:
> > On Mon, 19 Nov 2001, Peter Toneby wrote:
> > 
> > > On Mon, Nov 19, 2001 at 11:14:39AM -0500, Janina Sajka wrote:
> > > > What about freeamp?
> > > 
> > > It does not support searches on timecode (at least I didn't see anything
> > > about that while checking their pages, they only seem to have normal
> > > seek in files, which is not enough).
> > 
> > Well... It is enough with some extra processing. You need to find out the
> > data bitrate, estimate the seek position in the audio file, resynchronize on
> > a mpeg frame boundary, back a couple frames, compute the backed frames to
> > restabilize the mpeg windowing factors without actually playing them then
> > finally resume audio from the desired frame.
> 
> The problem is VBR, I'm not sure how it works, but I suspect each frame
> have it's own bitrate in that case, and if the stream use that, you need
> to loop through all frames and calculate the current time :(, well, I'll
> have a closer look at mad and see if I can do something for fixed
> bitrates at least.

No one can expect to use VBR in a talking book format and hope for snappy 
random seeks in the audio unless the indexing format now includes hard file 
offsets in bytes.  Last time I had to deal with that it wasn't the case and 
therefore VBR wasn't used either.


Nicolas