Hello Benny, As mentioned it is related to setting the file position. Our code always using pjsua_player_set_pos(), because that code can also be called for continueing playing a prompt where it was stopped. So after creating the wav_player port, you need to call the set_pos method. How to reproduce: 1. Create the player port (using the file I have provided which is clearly audible) 2. Next call next method: pjsua_player_set_pos() with offset is 0 3. Connect to the conference bridge to start playing. Action 2 will actually initiate the problem. Explained: In wav_player.c: If you check the fill_buffer(), you shall see that after reading from the file, the fport->data_left (represents the audio-data in chain) is decreased and the fpos (file position) is increased with the amount of data read (size). The fill_buffer() is called first time on the pjmedia_wav_player_port_create(). After this (within fill_buffer()), the file position is set +4000 bytes further and data_left is decreased with +4000. Now if you do the pjsua_player_set_pos(), with offset 0 (so re-start from beginning). This will call the pjmedia_wav_player_port_set_pos(), and the file position (fpos) will be reset. However.... fport->data_left is un-touched and still counts +4000 bytes. The last action in pjmedia_wav_player_port_set_pos() is again the fill_buffer(). >From this point there is a mis-match between fpos and data_left. As result the fill_buffer stops much earlier (because it is using ->data_left). So the more time you use _set_pos, the more mismatch there will be. Hope this helps. I've already deployed this patch and was told by our testers that it now works as expected. With regards, Eize > Hello Benny, > > > > I probably have the solution for mentioned problem. It was indeed that > the > > fport->data_left wasn't updated on a file-position change. > > Please have a check if I did not oversee something, and please add it to > the > > 1.x and 2.x tree. > > > > I just tried your WAV file with pjsua from the latest 1.x and 2.0, and > it sounds fine to me, with or without playback loop. Can you tell me > the exact way to reproduce it? > > Benny > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20111023/1f2575bf/attachment.html>
