My brother got married.

FFmpeg finally has channel layout support for AAC and Vorbis. I’ve published my work-in-progress HE AAC v1 (Spectral Band Replication) code in FFmpeg’s Summer of Code repository (though it has nothing to do with Google SoC).

A collaboration focused on open source software between two universities in Umeå, Sweden and Vaasa, Finland called OpenKvarken held a seminar about VoIP and related technologies. I did two talks there titled: An Open Standard IPTV Implementation in MythTV and Improving FFmpeg for High Definition Video Conferencing.

The latter refers to ffmpeg-mt – multi-threaded decoding for macroblock-based codecs as developed by Alexander Strange. OpenKvarken should be funding the merge of this code into FFmpeg trunk.

I should be getting a new job with Collabora working to improve the quality of open source multimedia on the whole. It’s a very exciting venture for me and it may lead to more regular blog posts.

SBR is coming along quite nicely and I hope to have it working soon.

I do have a pending patch to use some new IMDCT functions that Loren Merritt has introduced that take advantage of symmetries of the MDCT. The patch as it is has a bug in it that produces audible artifacts so I think I’ll need to draw out the different window types and check all the buffer indices. This isn’t a blocker though and I’ll maintain the AAC code after it’s committed to FFmpeg.

Loren suggested a rewrite of the overlapping and windowing code to output samples from the midpoint of the last MDCT block of the last frame to the midpoint of the last MDCT block of the current frame. This makes the code quite complex and may mess with post-processing style extensions like SBR so I don’t know if I’ll do this yet or not.

SBR is almost completely integrated and I’m starting debugging. Hopefully I’ll have that up and running within the next couple of weeks so Kostya can add SBR encoding to his AAC encoder he’s been working on for SoC.

Vladimir Voroshilov inspired me to briefly revisit my AMR-NB floating point encoder and we ironed out a number of bugs. We do at least have some sound output now, even if it is mostly garbage. I’ll jump back on that once I’m done with AAC or when I get sick of it for a moment.

The eventual goal will be to share some floating point functions in a similar fashion to Vladimir’s fixed point implementations of various speech codec routines. I expect writing a fixed point AMR-NB decoder wouldn’t be too difficult using Vladimir’s routines and I did also start work on this.

UPDATE: Please use the preset files from the 0.5 FFmpeg release or from a current subversion checkout of the FFmpeg development code. These preset files in this post are deprecated! As always, see the FFmpeg x264 encoding guide for up-to-date information.

I’ve uploaded a few libx264 ffpreset files but I’m still waiting for someone in the know to fix a bug in string option parsing for presets. The presets are:

• default – x264 CLI default values
• fastfirstpass – disables options unnecessary for first pass
• normal – Dark Shikari’s Q55 preset
• hq – Dark Shikari’s Q75 preset
• max – all options maxed out

If you dump them into ~/.ffmpeg/ then you can use them as noted in the previous post but due to the string option parsing bug you have to explicitly state the rc_eq option as follows. When this bug is fixed I’ll merge the rc_eq into the preset files.

ffmpeg -i infile -vcodec libx264 -vpre hq -rc_eq 'blurCplx(1-qComp)' -b 1M -bt 1M outfile.mp4

When I get bored again I’ll write some out for the different H.264 profiles and levels.

Have fun!

For example, one could have a preset file for high quality x264 encoding. The file would be ~/.ffmpeg/libx264-hq.ffpreset and would be used by `ffmpeg -i infile -vcodec libx264 -vpre hq -b 500k outfile.mp4` or something like that.

We have something in the works for per codec defaults that will greatly reduce the number of options required for x264 use in FFmpeg. Fun times in terms of usability. Needless to say I’ll be expending some effort making presets and getting them published somehow and also endeavouring to get useful per codec defaults applied as well.

The Google Summer of Code AAC decoder for FFmpeg is almost polished up. I have one patch pending at the time of writing and then I think I’ll be just about ready to request another full review. In the mean time I can start gluing in my SBR code and prodding it into life. Also fun times.