sön 2009-07-05 klockan 09:57 -0400 skrev Nicolas Dufresne: > Le dimanche 05 juillet 2009 à 14:37 +0200, Visti Andresen a écrit : > > I prefer using RAM and have the sound swapped out if memory is > > tight... > > Personally I would try other solution before going to the RAM solution. > The RAM solution will move the latency to your program startup time, > which is also a big issue. You could also decode and embed the sound in > a raw char* format but that will make the program less configurable (it > will also increase you program size, and thus slow down startup time). > > When you launch a sub-process, you need to load all the file to RAM > first, which is not the case for an audio file. For this reason, I would > suggest to test usinga streaming stack like GStreamer library. If all > your pipeline are brought to pause state (preroll and buffering done), > latency should be very low when time comes to play. You may get even > lower latency by enabling mmap() mode of your filesrc element. > > Best regards, > Nicolas If I can avoid C and stick to Erlang and the linux command line, eg. cat file.au /dev/audio, it would be a better solution. Erlang is like Java run in a virtual machine and like Java a JNI solution is not "pure". I can - just like in Java - compile a C function and use in Erlang as a normal function but it would not be portable. > > _______________________________________________ > devel mailing list > devel@lists.openmoko.org > https://lists.openmoko.org/mailman/listinfo/devel
signature.asc
Description: Detta är en digitalt signerad meddelandedel
_______________________________________________ devel mailing list devel@lists.openmoko.org https://lists.openmoko.org/mailman/listinfo/devel
