I would like to participate in GSoC this year and make some project for FFmpeg. As a qualification task I'm finishing last year's work - EAC3 decoder. This year I am keen to write some audio decoder. I'm interested in few ideas from the list. I will be pleased to hear any suggestions about them. G.723.1 Encoder: This task sounds good mainly because the documentations looks very good and there are decoder/encoder patterns. I've got one question: do you mean making fixed or floating-point encoder or both of them? ALAC Encoder: I would like to try to write the encoder base on alac.c and flac*.c, then try myself in efforts of developing existing decoder through reverse engineering. Nellymoser Encoder: I understand that here work should be based on the decoder code (I saw that the documentation is available, but it costs $6000), am I right? or possibly any other encoder is very necessary? (I'm keen to write it :) ) -- Bartlomiej Wolowiec
Bartlomiej Wolowiec wrote:
I would like to participate in GSoC this year and make some project for FFmpeg. As a qualification task I'm finishing last year's work - EAC3 decoder. This year I am keen to write some audio decoder. I'm interested in few ideas from the list. I will be pleased to hear any suggestions about them.
G.723.1 Encoder: This task sounds good mainly because the documentations looks very good and there are decoder/encoder patterns. I've got one question: do you mean making fixed or floating-point encoder or both of them?
Fixed point version is preferred, but a decoder is better then no decoder.
ALAC Encoder: I would like to try to write the encoder base on alac.c and flac*.c, then try myself in efforts of developing existing decoder through reverse engineering.
Nellymoser Encoder: I understand that here work should be based on the decoder code (I saw that the documentation is available, but it costs $6000), am I right?
The only reference is the decoder.
or possibly any other encoder is very necessary? (I'm keen to write it :) )
Then send in 3 applications for those project. MvH Benjamin Larsson
On poniedziałek, 31 marca 2008, Benjamin Larsson wrote:
Nellymoser Encoder: I understand that here work should be based on the decoder code (I saw that the documentation is available, but it costs $6000), am I right?
The only reference is the decoder.
I've got one more question: In this task (also in few other tasks) is mentioned about implementation of psychoacoustic model, so I searched for some information on this topic. I found http://wiki.multimedia.cx/index.php?title=Psychoacoustic_Model After a while of searching in FFmpeg source I haven't found a good psychoacoustic model implemented in any encoder. Did I miss something or maybe someone is working on it? Or maybe it's a good idea to use ready implementation from e.g. LAME and add more universal api with ability to use it in few encoders? -- Bartlomiej Wolowiec
Bartlomiej Wolowiec wrote:
On poniedziałek, 31 marca 2008, Benjamin Larsson wrote:
Nellymoser Encoder: I understand that here work should be based on the decoder code (I saw that the documentation is available, but it costs $6000), am I right? The only reference is the decoder.
I've got one more question: In this task (also in few other tasks) is mentioned about implementation of psychoacoustic model, so I searched for some information on this topic. I found http://wiki.multimedia.cx/index.php?title=Psychoacoustic_Model After a while of searching in FFmpeg source I haven't found a good psychoacoustic model implemented in any encoder. Did I miss something or maybe someone is working on it? Or maybe it's a good idea to use ready implementation from e.g. LAME and add more universal api with ability to use it in few encoders?
We'll the thing with Nellymoser is that the bitrate is really low, so the most crude psy-model would be to just cut the high frequencys and I'm not sure it is possible to do anything better then that. One would see after the bitstream encoder is done. MvH Benjamin Larsson
participants (2)
-
Bartlomiej Wolowiec -
Benjamin Larsson