[FFmpeg-devel] [PATCH] nellymoserdec: Allow using unusual input block sizes

Martin Storsjö martin
Mon Jun 14 14:38:12 CEST 2010

On Mon, 14 Jun 2010, Benjamin Larsson wrote:

> Martin Storsj? skrev 2010-06-14 13:27:
> > 
> > Currently, the nellymoser decoder tries to decode all blocks in a packet
> > at once (for efficiency?), instead of only decoding the first one (as many
> > other decoders do). The audio decoders that work in that way are able to
> > decode a long stream of data without any framing, as long as they're fed
> > with enough data to decode one packet.
> > 
> > If the amount of blocks in the packet isn't one of the ones the decoder is
> > familiar with, it simply errors out (with a message at the AV_LOG_DEBUG
> > level, btw) - even if it would have decoded just fine if it would decode
> > only one frame at a time.
> > 
> > The attached patch is one way of solving it, another (less efficient) way
> > would be to only decode the first block of a packet per call.
> > 
> Patch ok, with a following indentation commit.

Applied, thanks for the quick reply.

Do you think it would be a good idea to increase the log level of the 
message to AV_LOG_ERROR, too? At the moment, it's not too easy to realize 
why the decoder doesn't output anything.

// Martin

More information about the ffmpeg-devel mailing list