Re: Using available DSP tasks
Quoting Daniel Charles [EMAIL PROTECTED]: See the merged pipeline below Yes, this is certainly doable already. I don't have any G7.11 data to hand, but have tried it by mixing ogg and mp3 data (ogg using Tuomas Kulve's gstreamer plugin which uses the pcm dsp sink). E.g. These two commands can be run is separate terminal windows and are mixed: gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink gst-launch-0.10 filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink \ filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink This way you can be sure that both pipelines are running on a single process. I'm not certain it is going to work as expected due to constraints with two audio streams running at the same time, but you can write as many source-filter-sink pipelines in a single gst-launch as you want. Thank you very much. Is the backslash optional? Was it there just to indicate a line continuation (my email client has wrapped the following line too) or is it the correct way to separate the different parts of the pipeline? In any case your command (with or without backslash) produces the desired behaviour and both files play. Cheers, Simon P.S. Just to test that the karaoke idea will work I tested MP3 and G7.29 data (I couldn't find any G7.11 data to test) and they play together without troubles: E.g. gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink filesrc location=./audio-temp/transfer.g729 ! dspg729sink ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
On Jan 1, 2008 12:32 PM, Simon Pickering [EMAIL PROTECTED] wrote: Quoting Daniel Charles [EMAIL PROTECTED]: See the merged pipeline below Yes, this is certainly doable already. I don't have any G7.11 data to hand, but have tried it by mixing ogg and mp3 data (ogg using Tuomas Kulve's gstreamer plugin which uses the pcm dsp sink). E.g. These two commands can be run is separate terminal windows and are mixed: gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink gst-launch-0.10 filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink \ filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink This way you can be sure that both pipelines are running on a single process. I'm not certain it is going to work as expected due to constraints with two audio streams running at the same time, but you can write as many source-filter-sink pipelines in a single gst-launch as you want. Thank you very much. Is the backslash optional? Was it there just to indicate a line continuation (my email client has wrapped the following line too) or is it the correct way to separate the different parts of the pipeline? Yes it is a line continuation, normally in a command prompt you can either type (copy/paste) the entire line or with the back slash for read clarity but it is optional. In any case your command (with or without backslash) produces the desired behaviour and both files play. I'm glad to hear that it worked :) Daniel Cheers, Simon P.S. Just to test that the karaoke idea will work I tested MP3 and G7.29 data (I couldn't find any G7.11 data to test) and they play together without troubles: E.g. gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink filesrc location=./audio-temp/transfer.g729 ! dspg729sink ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Hello Simon and Daniel, Your help is truly fantastic! I have found the gst-inspect-0.10 -a command and I have now almost all the informations I need. I have tested this command: gst-launch-0.10 dspg729src dtx=3 ! dspg729sink filesrc location=MyDocs/.sound/Here_is_the_story.mp3 Most of the time this command raise a error in gst_dspg729src_start: could not open resource for reading and writing. But sometimes it work perfectly well as expected, with as bonus the automatic gain control of the voice with silence detection, and all that without loading the main CPU. Very nice! But the following command never work as I expected: gst-launch-0.10 dspg729src dtx=3 ! filesink location=in filesrc location=out ! dspg729sink filesrc location=MyDocs/.sound/Here_is_the_story.mp3 The in file start recorded only after the out file has been played. And sometimes the command failed with an error in gst_dspmp3sink_open : Could not open resource for reading and writing, witch is strange for the MP3 stream. Other problem, the G729 stream is only usable for a few seconds if it is recorded into a file. I actually don't know if the problem is while recording or playing the stream. I suspect that I maybe need to use something that look more like stream of packets instead of a file to handle the G729 stream. Best Regards, -- Jean-Christian P.S. Just to test that the karaoke idea will work I tested MP3 and G7.29 data (I couldn't find any G7.11 data to test) and they play together without troubles: E.g. gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink filesrc location=./audio-temp/transfer.g729 ! dspg729sink ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Sorry, I missed the ! dspmp3sink at the end of each command into my previous mail. The correct commands was: gst-launch-0.10 dspg729src dtx=3 ! dspg729sink filesrc location=MyDocs/.sound/Here_is_the_story.mp3 ! dspmp3sink gst-launch-0.10 dspg729src dtx=3 ! filesink location=in filesrc location=out ! dspg729sink filesrc location=MyDocs/.sound/Here_is_the_story.mp3 ! dspmp3sink -- Jean-Christian de Rivaz ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Thanks for your explanation. np After having read the basic Gstreamer documentation, I understand better the sink pad concept of the mp3 task. In the application I am thinking about now, I don't need to look at the raw audio data decoded by the MP3 task as long as I can mix with it an other raw audio stream. I fact I need to mix a MP3 stream and a G711|G729|ILBC|AMR stream. You can see it as a kind of karaoke: mixing music (MP3) and voice (low bandwidth codec). The result should simply be available on the output jack. Did you think it is already doable now ? Yes, this is certainly doable already. I don't have any G7.11 data to hand, but have tried it by mixing ogg and mp3 data (ogg using Tuomas Kulve's gstreamer plugin which uses the pcm dsp sink). E.g. These two commands can be run is separate terminal windows and are mixed: gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink gst-launch-0.10 filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink So this uses the mp3 sink and the pcm sink simultaneously. I imagine it must be possible to start two separate sources in a single gstreamer pipeline and then send them to separate sinks (i.e. as I've done in the above example but all in one command). Anyone know how to do this? Simon ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Hi, See the merged pipeline below On Dec 31, 2007 7:49 AM, Simon Pickering [EMAIL PROTECTED] wrote: Thanks for your explanation. np After having read the basic Gstreamer documentation, I understand better the sink pad concept of the mp3 task. In the application I am thinking about now, I don't need to look at the raw audio data decoded by the MP3 task as long as I can mix with it an other raw audio stream. I fact I need to mix a MP3 stream and a G711|G729|ILBC|AMR stream. You can see it as a kind of karaoke: mixing music (MP3) and voice (low bandwidth codec). The result should simply be available on the output jack. Did you think it is already doable now ? Yes, this is certainly doable already. I don't have any G7.11 data to hand, but have tried it by mixing ogg and mp3 data (ogg using Tuomas Kulve's gstreamer plugin which uses the pcm dsp sink). E.g. These two commands can be run is separate terminal windows and are mixed: gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink gst-launch-0.10 filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink gst-launch-0.10 filesrc location=20-16000HzExp5sec.mp3 ! dspmp3sink \ filesrc location=opensource.ogg ! application/ogg ! tremor ! dsppcmsink This way you can be sure that both pipelines are running on a single process. I'm not certain it is going to work as expected due to constraints with two audio streams running at the same time, but you can write as many source-filter-sink pipelines in a single gst-launch as you want. Thanks, Daniel. So this uses the mp3 sink and the pcm sink simultaneously. I imagine it must be possible to start two separate sources in a single gstreamer pipeline and then send them to separate sinks (i.e. as I've done in the above example but all in one command). Anyone know how to do this? Simon ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Jean-Christian de Rivaz wrote: g729_dec was runnings. I wonder if this was a limitation from the DSP or from the Linux applications (e.g. Media Player and Skype trying to use the same device). I think the device stops all other audio when a VOIP call is made, but I think this is how it's designed, not a technical limitation. I'm mostly guessing here though. You should be able to test that quite easily by running couple gst-launch instances from the command line. -- Tuomas signature.asc Description: OpenPGP digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
If the DSP in not the limitation, then the N800 could be used for a funny project I have now in my head. I there any documentation on how to program a Linux application to use some existing DSP tasks available on the N800? I am interesting about the MP3 decoder and a pair of low bandwidth coder/decoder like G711, G729 or ILBC. There is source available for the ARM-side part of the gstreamer sinks (http://repository.maemo.org/pool/chinook/free/source/g/gst-plugins-dsp0.10/). That should show you how to use the dsp tasks. There's a fundamental problem though, the mp3 dsp task sends its data directly to the audio codec (hardware) on the DSP-side, therefore you're probably not going to be able to access the raw data to reencode it as something else (unless it is also exposed in a buffer in shared memory). I am debugging an mp3 dsptask at the moment, so you may yet have something to play with at some point in the near future if you can't get the built-in tasks to work as you wish. Cheers, Simon ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Simon Pickering wrote: There is source available for the ARM-side part of the gstreamer sinks (http://repository.maemo.org/pool/chinook/free/source/g/gst-plugins-dsp0.10/). That should show you how to use the dsp tasks. Source that is non-compilable: https://bugs.maemo.org/show_bug.cgi?id=2271 -- Tuomas signature.asc Description: OpenPGP digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Tuomas Kulve a écrit : Jean-Christian de Rivaz wrote: g729_dec was runnings. I wonder if this was a limitation from the DSP or from the Linux applications (e.g. Media Player and Skype trying to use the same device). I think the device stops all other audio when a VOIP call is made, but I think this is how it's designed, not a technical limitation. I'm mostly guessing here though. You should be able to test that quite easily by running couple gst-launch instances from the command line. Thanks for the gst-launch hint. I didn't know Gstreamer, so I will have to learn a bit before doing useful experiment. Seem to be a very nice open source framework. Regards, -- Jean-Christian de Rivaz ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Re: Using available DSP tasks
Simon Pickering a écrit : If the DSP in not the limitation, then the N800 could be used for a funny project I have now in my head. I there any documentation on how to program a Linux application to use some existing DSP tasks available on the N800? I am interesting about the MP3 decoder and a pair of low bandwidth coder/decoder like G711, G729 or ILBC. There is source available for the ARM-side part of the gstreamer sinks (http://repository.maemo.org/pool/chinook/free/source/g/gst-plugins-dsp0.10/). That should show you how to use the dsp tasks. There's a fundamental problem though, the mp3 dsp task sends its data directly to the audio codec (hardware) on the DSP-side, therefore you're probably not going to be able to access the raw data to reencode it as something else (unless it is also exposed in a buffer in shared memory). I am debugging an mp3 dsptask at the moment, so you may yet have something to play with at some point in the near future if you can't get the built-in tasks to work as you wish. Cheers, Hi Simon, Thanks for your explanation. After having read the basic Gstreamer documentation, I understand better the sink pad concept of the mp3 task. In the application I am thinking about now, I don't need to look at the raw audio data decoded by the MP3 task as long as I can mix with it an other raw audio stream. I fact I need to mix a MP3 stream and a G711|G729|ILBC|AMR stream. You can see it as a kind of karaoke: mixing music (MP3) and voice (low bandwidth codec). The result should simply be available on the output jack. Did you think it is already doable now ? Best Regards, -- Jean-Christian de Rivaz ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
Using available DSP tasks
Hello, I have monitored the DPS tasks by executing periodically the command cat /sys/devices/platform/dsp/dsptask*/taskname while playing some tests with a MP3 in media player and the Skype test call. I found that the DSP can run many tasks at the same time, but I was unable to catch the mp3dec task running at the same time that the g729_enc and g729_dec was runnings. I wonder if this was a limitation from the DSP or from the Linux applications (e.g. Media Player and Skype trying to use the same device). If the DSP in not the limitation, then the N800 could be used for a funny project I have now in my head. I there any documentation on how to program a Linux application to use some existing DSP tasks available on the N800? I am interesting about the MP3 decoder and a pair of low bandwidth coder/decoder like G711, G729 or ILBC. Best Regards, -- Jean-Christian de Rivaz ___ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
