Bartlomiej Wolowiec wrote: > Currently, I'm trying to make changes which allow opening files with more > channels than 5.1. Now the decoder is written in a way, which allows > ac3_decode_frame to decode one frame, but additional channels in E-AC3 are > sent in separate frames. Due to this, I have a question, how to solve it: > - ac3_decode_frame function should decode many frames to get complete > information about outstream, or > - ac3_decode_frame function should return outstream data when it is called > for > the first frame from the next data package?
I definitely agree with Michael that the parser should send multiple AC3 frames as a single unit to the decoder. So the decoder will have to essentially process 2 (or more) frames at the same time. My suggestion is to break up the decode context since you will have to keep state information for both frames. Because each block is processed sequentially, the decoding process would need to be split up as well so that you can decode the transform coefficients from all AC3 frames in the set for a single block before downmixing. I understand that this work is a continuation of last year, but because the decoder is getting to a point where we want to add it to FFmpeg SVN soon, please post a patch of your solution to this before committing it. Thanks, Justin _______________________________________________ FFmpeg-soc mailing list [email protected] https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-soc
