Trick modes in GStreamer

Jan just demoed trick modes on his machine using GStreamer. Trick modes in the term used for stuff like double speed, quadruple speed, half speed, backwards play etc. playback. A little code cleanup is still needed before comitting, but it will be in GStreamer CVS next week.

The initial goal is supporting server side trickmode properly like when getting a video feed from a ViiV enabled server, but it already supports some modes of client side trickmodes too.

Along with the recently added Quality of Service framework and network clocks we are adding a lot of advanced functionality to GStreamer these days.

Edward and Wim also did some critical fixes in GStreamer and GNonlin today for Jokosher, to help ensure that the Jokosher team will be able to demo a working application at GUADEC.

Google Summer of code deadline is approaching fast (monday morning) and we still have room for more students to propose GStreamer related projects under the GNOME, Xiph or BBC banners. Be sure to check out these projects ideas lists and submit a proposal. Probably other projects also stil open to more proposals, but I am not mentoring those so I don’t know their status.


#1 Anonymous on 05.07.06 at 03:27

Thank you! This feature alone has kept me occasionally using MPlayer rather than Totem. I look forward to hitting a key in Totem to speed up or slow down playback.

One question: do you just speed up or slow down the audio as well, like MPlayer does, or do you change the audio speed without changing the pitch, like MythTV does? I’d really love to see the latter, though even the former will help.

#2 Anonymous on 05.07.06 at 18:25

I can see GStreamer trick modes being useful for playing back language audio, for instance, when slowing down “tape speech” is desired.

With .edu being a big target for Gnome and Linux, I wonder if there could also be an alternative UI (mode) specifically aimed at playing back language study audio etc., with easy ways of adjusting playback speed, length of breaks, the number of repeats and even letting the listener to record his/her own pronunciation after each track and compare the two.

The other side of the equation would then be authoring tools allowing educators to easily create organized and labelled audio and video oggs.

And letting people take 45 min lectures (like “albums”) and tag them with labels or headers denoting virtual tracks which could then be played back desired number of times in desired order etc.

OK, these features would probably be up to other frontends to GStreamer but anyway, someone could help revolutionize the often passive way languages are studied.