[alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
I am a newbie to ALSA and any help will be much appreciated.
I have an application which sets up the signaling between an IP phone and my desktop and then sets up the audio path between the two.
On my desktop application, I am able to receive RTP packets from IP phone. I use an RTP stack to parse the data and after going through the RTP stack, I try to play back the audio using ALSA. When I use the ALSA code to play back (in real time) using sound card of my device, there is only noise, I cannot hear the audio that I speak into the IP phone. However, if I take the raw data coming from the RTP stack and write it to a file, I can play back the file successfully.
Since my data from IP phone is G.711 encoded, I have set the sampling rate within ALSA to 8000. Also I am using one source (mono) and non-interleaved data option for preparing ALSA for playback. When I set the format to SND_PCM_FORMAT_MU_LAW, at runtime it lets me set that format ie. snd_pcm_hw_params_set_format (for SND_PCM_FORMAT_MU_LAW) is successful. However I get a runtime error while applying the hardware parameters using snd_pcm_hw_params(..) if the format set earlier is SND_PCM_FORMAT_MU_LAW. Using any other format like SND_PCM_FORMAT_U8 or SND_PCM_FORMAT_S8, lets me apply the hardware parameters but it gives the problem of the noise (no "audible" voice) that I described earlier.
This is what my ALSA code looks like
***** beginning of code snippet *****************
int err;
snd_pcm_hw_params_t *hw_params;
// Try to open the default device
err = snd_pcm_open( &_soundDevice, "plughw:0,0", SND_PCM_STREAM_PLAYBACK, 0 );
// Check for error on open.
if( err < 0 )
{
printf("Init: cannot open audio device\n");
return -1;
}
// Allocate the hardware parameter structure.
if ((err = snd_pcm_hw_params_malloc (&hw_params)) < 0)
{
printf("Init: cannot allocate hardware parameter structure\n");
return -1;
}
if ((err = snd_pcm_hw_params_any (_soundDevice, hw_params)) < 0)
{
printf("Init: cannot initialize hardware parameter structure\n");
return -1;
}
// Set access to RW interleaved.
if ((err = snd_pcm_hw_params_set_access (_soundDevice, hw_params, SND_PCM_ACCESS_RW_NONINTERLEAVED)) < 0)
{
printf("Init: cannot set access type\n");
return -1;
}
if ((err = snd_pcm_hw_params_set_format (_soundDevice, hw_params, SND_PCM_FORMAT_MU_LAW)) < 0)
{
return -1;
}
// Set channels to mono (1)
if ((err = snd_pcm_hw_params_set_channels (_soundDevice, hw_params, 1)) < 0)
{
return -1;
}
// Set sample rate.
unsigned int actualRate = 8000;
if ((err = snd_pcm_hw_params_set_rate_near (_soundDevice, hw_params, &actualRate,
0)) < 0)
{
return -1;
}
// Apply the hardware parameters that we've set.
if ((err = snd_pcm_hw_params (_soundDevice, hw_params)) < 0)
{
printf("Audio device parameters have not been set successfully.(%s)(%s)\n", strerror (err), snd_strerror (err));
return -1;
}
// Free the hardware parameters now that we're done with them.
snd_pcm_hw_params_free (hw_params);
/* Prepare interface for use.
if ((err = snd_pcm_prepare (_soundDevice)) < 0)
{
printf("Audio device not prepared for use\n");
return -1;
}
// If the frames are received...
/* find out how much space is available for playback data
if ((frames_to_deliver = snd_pcm_avail_update (_soundDevice)) < 0)
{
printf("Error returned by snd_pcm_avail_update\n");
}
frames_to_deliver = frames_to_deliver > 4096 ? 4096 : frames_to_deliver;
printf("frames to deliver is %d\n",(int)frames_to_deliver);
*/
frames_to_deliver = 172;
bufs[1] = (void*)buffer;
if ((err = snd_pcm_writen( _soundDevice, bufs, frames_to_deliver)) < 0)
{
printf ("write failed (%s)\n", snd_strerror (err));
}
****************** end of code snippet **********************************
The error I get is as follows:-
Audio device parameters have not been set successfully.(Unknown error 4294967274)(Invalid argument)
Any pointers on what I could be doing wrong?
Thanks and regards,
Mitul
The code is correct (as intended). I am sorry I forgot to change the comment. I think I should be using SND_PCM_ACCESS_RW_NONINTERLEAVED since I have only one source (mono). Please correct me if that is wrong.
Regards, Mitul
-----Original Message----- From: alsa-devel-bounces@alsa-project.org [mailto:alsa-devel-bounces@alsa-project.org] On Behalf Of Clemens Ladisch Sent: Tuesday, July 01, 2008 12:39 AM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
// Set access to RW interleaved. if ((err = snd_pcm_hw_params_set_access (_soundDevice, hw_params, SND_PCM_ACCESS_RW_NONINTERLEAVED)) < 0)
The comment and the code do not agree. Please try SND_PCM_ACCESS_RW_INTERLEAVED.
Regards, Clemens _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
Mitul Sen (misen) wrote:
I am a newbie to ALSA and any help will be much appreciated.
I have an application which sets up the signaling between an IP phone and my desktop and then sets up the audio path between the two.
On my desktop application, I am able to receive RTP packets from IP phone. I use an RTP stack to parse the data and after going through the RTP stack, I try to play back the audio using ALSA. When I use the ALSA code to play back (in real time) using sound card of my device, there is only noise, I cannot hear the audio that I speak into the IP phone. However, if I take the raw data coming from the RTP stack and write it to a file, I can play back the file successfully.
Since my data from IP phone is G.711 encoded, I have set the sampling rate within ALSA to 8000. Also I am using one source (mono) and non-interleaved data option for preparing ALSA for playback. When I set the format to SND_PCM_FORMAT_MU_LAW, at runtime it lets me set that format ie. snd_pcm_hw_params_set_format (for SND_PCM_FORMAT_MU_LAW) is successful. However I get a runtime error while applying the hardware parameters using snd_pcm_hw_params(..) if the format set earlier is SND_PCM_FORMAT_MU_LAW. Using any other format like SND_PCM_FORMAT_U8 or SND_PCM_FORMAT_S8, lets me apply the hardware parameters but it gives the problem of the noise (no "audible" voice) that I described earlier.
This sounds like there is a mismatch in the data. Two suggestions.
Look at the code that implements SND_PCM_FORMAT_MU_LAW in alsa-lib and make sure it is correct (or actually there and not a stub). See if there is some other parameter(s) you have to set in order for it to function.
Write the output being sent to alsa-lib to a file as you are sending it and compare it to the raw data you have captured that plays OK. You want it to be the same. If it isn't, that is your problem. If it is, the problem is in the alsa-lib decoding for this case. Obviously, the coding must work in some case, because you are playing the raw data successfully. Either the application or alsa is decoding it successfully. Compare the successful decoding process with the unsuccessful decoding process.
Thanks for your suggestions. For the first suggestion, how would I know the version of alsa-lib that I am using so that I can download the source code for the same? Is alsa-lib the same as libasound? Under /usr/lib/libasound.so I have a symbolic link to libasound.so.2.0.0. Is that then the version I should be looking for?
I am really new to this so please excuse me if these questions are too basic for this forum!
Thanks, Mitul
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Tue 7/1/2008 6:59 AM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
I am a newbie to ALSA and any help will be much appreciated.
I have an application which sets up the signaling between an IP phone and my desktop and then sets up the audio path between the two.
On my desktop application, I am able to receive RTP packets from IP phone. I use an RTP stack to parse the data and after going through the RTP stack, I try to play back the audio using ALSA. When I use the ALSA code to play back (in real time) using sound card of my device, there is only noise, I cannot hear the audio that I speak into the IP phone. However, if I take the raw data coming from the RTP stack and write it to a file, I can play back the file successfully.
Since my data from IP phone is G.711 encoded, I have set the sampling rate within ALSA to 8000. Also I am using one source (mono) and non-interleaved data option for preparing ALSA for playback. When I set the format to SND_PCM_FORMAT_MU_LAW, at runtime it lets me set that format ie. snd_pcm_hw_params_set_format (for SND_PCM_FORMAT_MU_LAW) is successful. However I get a runtime error while applying the hardware parameters using snd_pcm_hw_params(..) if the format set earlier is SND_PCM_FORMAT_MU_LAW. Using any other format like SND_PCM_FORMAT_U8 or SND_PCM_FORMAT_S8, lets me apply the hardware parameters but it gives the problem of the noise (no "audible" voice) that I described earlier.
This sounds like there is a mismatch in the data. Two suggestions.
Look at the code that implements SND_PCM_FORMAT_MU_LAW in alsa-lib and make sure it is correct (or actually there and not a stub). See if there is some other parameter(s) you have to set in order for it to function.
Write the output being sent to alsa-lib to a file as you are sending it and compare it to the raw data you have captured that plays OK. You want it to be the same. If it isn't, that is your problem. If it is, the problem is in the alsa-lib decoding for this case. Obviously, the coding must work in some case, because you are playing the raw data successfully. Either the application or alsa is decoding it successfully. Compare the successful decoding process with the unsuccessful decoding process.
Mitul Sen (misen) wrote:
Thanks for your suggestions. For the first suggestion, how would I know the version of alsa-lib that I am using so that I can download the source code for the same? Is alsa-lib the same as libasound? Under /usr/lib/libasound.so I have a symbolic link to libasound.so.2.0.0. Is that then the version I should be looking for?
I am really new to this so please excuse me if these questions are too basic for this forum!
Thanks, Mitul
Good queston. I know because of the package. You can find the driver version by running cat /proc/asound/version and the utilities version by running aplay --version . I couldn't find a way to get the library version. You can use the run alsa-info.sh --no-upload and the output will have the driver, library, and tool version at the top. The script is found at http://www.alsa-project.org/alsa-info.sh
libasound is always version 2.0.0, no matter the alsa-lib version. I presume that is to facilitate testing and upgrading. Nothing will break with newer versions. So that is unrelated to alsa version.
You can find the latest beta versions of all the alsa source here
http://www.alsa-project.org/main/index.php/Changes_v1.0.17rc2_v1.0.17rc3
And a link to the latest stable release here http://www.alsa-project.org
Hi Stan,
Thanks for all your help! I have some more questions though...
I downloaded the source code for alsa-lib-1.0.15 Based on the code, if the format is SND_PCM_FORMAT_MU_LAW, I am not sure why it does a get/put index to SND_PCM_FORMAT_S16 Also, if the stream is SND_PCM_STREAM_PLAYBACK, then I would think that it should decode the data. Why does it call snd_pcm_mulaw_decode function if the format is SND_PCM_FORMAT_MU_LAW and snd_pcm_mulaw_encode otherwise. I have an Intel HDA soundcard and according to the specs, it should support PCM ulaw format.
All ALSA documentation and examples I have come across use specific hw_params (like sample rate of 44100, SND_PCM_FORMAT_S16, 2 channel interleaved data). According to the documents, hw_params refer to the stream related info so that's the reason I tried to change it to that of mu-law (sampling rate of 8000 Hz, SND_PCM_FORMAT_MU_LAW etc). Not sure if that's the way to do it though. Based on the code it looks like the hardware just seems to support SND_PCM_FORMAT_S16. Any pointers to help me better understand the ALSA code would be much appreciated.
The code that I am referring to is in pcm_mulaw.c and is as follows:-
static int snd_pcm_mulaw_hw_params(snd_pcm_t *pcm, snd_pcm_hw_params_t * params) { snd_pcm_mulaw_t *mulaw = pcm->private_data; snd_pcm_format_t format; int err = snd_pcm_hw_params_slave(pcm, params,
snd_pcm_mulaw_hw_refine_cchange,
snd_pcm_mulaw_hw_refine_sprepare,
snd_pcm_mulaw_hw_refine_schange, snd_pcm_generic_hw_params); if (err < 0) return err;
err = INTERNAL(snd_pcm_hw_params_get_format)(params, &format); if (err < 0) return err;
if (pcm->stream == SND_PCM_STREAM_PLAYBACK) { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx = snd_pcm_linear_get_index(format, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } else { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, mulaw->sformat); mulaw->func = snd_pcm_mulaw_decode; } } else { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, format); mulaw->func = snd_pcm_mulaw_decode; } else { mulaw->getput_idx = snd_pcm_linear_get_index(mulaw->sformat, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } } return 0; }
Thanks and regards, Mitul
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Tuesday, July 01, 2008 3:12 PM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
Thanks for your suggestions. For the first suggestion, how would I know the version of alsa-lib that I am using so that I can download the source code for the same? Is alsa-lib the same as libasound? Under
/usr/lib/libasound.so I have a symbolic link to libasound.so.2.0.0. Is
that then the version I should be looking for?
I am really new to this so please excuse me if these questions are too
basic for this forum!
Thanks, Mitul
Good queston. I know because of the package. You can find the driver version by running cat /proc/asound/version and the utilities version by running aplay --version . I couldn't find a way to get the library version. You can use the run alsa-info.sh --no-upload and the output will have the driver, library, and tool version at the top. The script is found at http://www.alsa-project.org/alsa-info.sh
libasound is always version 2.0.0, no matter the alsa-lib version. I presume that is to facilitate testing and upgrading. Nothing will break with newer versions. So that is unrelated to alsa version.
You can find the latest beta versions of all the alsa source here
http://www.alsa-project.org/main/index.php/Changes_v1.0.17rc2_v1.0.17rc3
And a link to the latest stable release here http://www.alsa-project.org
Mitul Sen (misen) wrote:
Hi Stan,
Thanks for all your help! I have some more questions though...
I downloaded the source code for alsa-lib-1.0.15 Based on the code, if the format is SND_PCM_FORMAT_MU_LAW, I am not sure why it does a get/put index to SND_PCM_FORMAT_S16 Also, if the stream is SND_PCM_STREAM_PLAYBACK, then I would think that it should decode the data. Why does it call snd_pcm_mulaw_decode function if the format is SND_PCM_FORMAT_MU_LAW and snd_pcm_mulaw_encode otherwise. I have an Intel HDA soundcard and according to the specs, it should support PCM ulaw format.
All ALSA documentation and examples I have come across use specific hw_params (like sample rate of 44100, SND_PCM_FORMAT_S16, 2 channel interleaved data). According to the documents, hw_params refer to the stream related info so that's the reason I tried to change it to that of mu-law (sampling rate of 8000 Hz, SND_PCM_FORMAT_MU_LAW etc). Not sure if that's the way to do it though. Based on the code it looks like the hardware just seems to support SND_PCM_FORMAT_S16. Any pointers to help me better understand the ALSA code would be much appreciated.
Hi Misen,
First, a gentle remonstrance. You probably have noticed that I always put my responses after or mixed with your message. On public mailing lists this is considered good form, rather than posting your response at the top of the message. Why? So that anyone who steps into the interaction doesn't have to read the messages out of order and that future searchers have an easier time understanding the message. While top posting is the norm in communications between two or a few people because the context is familiar to all and it saves time not to have to look for the response, on a public mailing list that isn't necessarily true.
Now to the matter at hand. I had never heard of mu law so I looked it up. http://www.digitalpreservation.gov/formats/fdd/fdd000039.shtml ... Standard companding algorithm used in digital communications systems in North America and Japan (telephones, for the most part) to optimize the dynamic range of an analog signal (generally a voice) for digitizing, i.e., to compress 16 bit LPCM http://www.digitalpreservation.gov/formats/fdd/fdd000011.shtml (Linear Pulse Code Modulated) data down to 8 bits of logarithmic data. See also Notes http://www.digitalpreservation.gov/formats/fdd/fdd000039.shtml#notes below. µ-Law is similar to the A-Law http://www.digitalpreservation.gov/formats/fdd/fdd000038.shtml algorithm used in Europe. ...
The code that you extracted below is designed to convert mu law from the compressed form back into the 16 bit signed form. I haven't checked the rest of the code myself, but it appears to assume that the sound device is incapable of internal conversion. If that is true, you shouldn't have to specify anything else to the library except mu law. It should take care of everything else. i.e. as soon as you specify mu law, it is known that the stream is 8 bit mono that has to be uncompressed to 16 bit mono. I presume that is why there is the error when you try to set the hardware parms with mu law. The library should probably be modified to use this new capability of sound device internal conversion for mu law if it is available on the sound device. Maybe it already does; as I said I haven't looked at the code, and I'm not really familiar with mu law.
So, given my ignorance, my explanation and proposed solution might be completely wrong. :-) Perhaps a developer familiar with the coding of mu law will give a better explanation.
At this point, I really don't have more to offer for your problem. I would have to look at the code to decipher it in order to give an answer. You might as well do that yourself, as you will get a better understanding than I could give with an explanation.
The code that I am referring to is in pcm_mulaw.c and is as follows:-
static int snd_pcm_mulaw_hw_params(snd_pcm_t *pcm, snd_pcm_hw_params_t * params) { snd_pcm_mulaw_t *mulaw = pcm->private_data; snd_pcm_format_t format; int err = snd_pcm_hw_params_slave(pcm, params,
snd_pcm_mulaw_hw_refine_cchange,
snd_pcm_mulaw_hw_refine_sprepare,
snd_pcm_mulaw_hw_refine_schange, snd_pcm_generic_hw_params); if (err < 0) return err;
err = INTERNAL(snd_pcm_hw_params_get_format)(params, &format); if (err < 0) return err; if (pcm->stream == SND_PCM_STREAM_PLAYBACK) { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx =snd_pcm_linear_get_index(format, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } else { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, mulaw->sformat); mulaw->func = snd_pcm_mulaw_decode; } } else { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, format); mulaw->func = snd_pcm_mulaw_decode; } else { mulaw->getput_idx = snd_pcm_linear_get_index(mulaw->sformat, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } } return 0; }
Thanks and regards, Mitul
-----Or
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Tuesday, July 01, 2008 7:20 PM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
Hi Stan,
Thanks for all your help! I have some more questions though...
I downloaded the source code for alsa-lib-1.0.15 Based on
the code, if
the format is SND_PCM_FORMAT_MU_LAW, I am not sure why it does a get/put index to SND_PCM_FORMAT_S16 Also, if the stream is SND_PCM_STREAM_PLAYBACK, then I would think that it should
decode the
data. Why does it call snd_pcm_mulaw_decode function if the
format is
SND_PCM_FORMAT_MU_LAW and snd_pcm_mulaw_encode otherwise. I have an Intel HDA soundcard and according to the specs, it should
support PCM
ulaw format.
All ALSA documentation and examples I have come across use specific hw_params (like sample rate of 44100, SND_PCM_FORMAT_S16, 2 channel interleaved data). According to the documents, hw_params
refer to the
stream related info so that's the reason I tried to change
it to that
of mu-law (sampling rate of 8000 Hz, SND_PCM_FORMAT_MU_LAW
etc). Not
sure if that's the way to do it though. Based on the code it looks like the hardware just seems to support SND_PCM_FORMAT_S16. Any pointers to help me better understand the ALSA code would
be much appreciated.
Hi Misen,
First, a gentle remonstrance. You probably have noticed that I always put my responses after or mixed with your message. On public mailing lists this is considered good form, rather than posting your response at the top of the message. Why? So that anyone who steps into the interaction doesn't have to read the messages out of order and that future searchers have an easier time understanding the message. While top posting is the norm in communications between two or a few people because the context is familiar to all and it saves time not to have to look for the response, on a public mailing list that isn't necessarily true.
Point noted!
Now to the matter at hand. I had never heard of mu law so I looked it up. http://www.digitalpreservation.gov/formats/fdd/fdd000039.shtml ... Standard companding algorithm used in digital communications systems in North America and Japan (telephones, for the most part) to optimize the dynamic range of an analog signal (generally a voice) for digitizing, i.e., to compress 16 bit LPCM http://www.digitalpreservation.gov/formats/fdd/fdd000011.shtm l (Linear Pulse Code Modulated) data down to 8 bits of logarithmic data. See also Notes http://www.digitalpreservation.gov/formats/fdd/fdd000039.shtml#notes below. µ-Law is similar to the A-Law http://www.digitalpreservation.gov/formats/fdd/fdd000038.shtml algorithm used in Europe. ...
The code that you extracted below is designed to convert mu law from the compressed form back into the 16 bit signed form. I haven't checked the rest of the code myself, but it appears to assume that the sound device is incapable of internal conversion. If that is true, you shouldn't have to specify anything else to the library except mu law. It should take care of everything else. i.e. as soon as you specify mu law, it is known that the stream is 8 bit mono that has to be uncompressed to 16 bit mono. I presume that is why there is the error when you try to set the hardware parms with mu law.
Thanks! You are right. I don't get the error when I only specify mu law and let the library convert it to 16 bit signed form. I can now hear "some" audio but it sounds very faint. I have been playing around with the frame size because I seem to be able to hear the audio only when I use a small frame size. I need to get a better understanding of the code and my sound device to improve the audio quality. For now, at least I am able to set the hardware parms with mulaw.
The library should probably be modified to use this new capability of sound device internal conversion for mu law if it is available on the sound device. Maybe it already does; as I said I haven't looked at the code, and I'm not really familiar with mu law.
So, given my ignorance, my explanation and proposed solution might be completely wrong. :-) Perhaps a developer familiar with the coding of mu law will give a better explanation.
At this point, I really don't have more to offer for your problem. I would have to look at the code to decipher it in order to give an answer. You might as well do that yourself, as you will get a better understanding than I could give with an explanation.
The code that I am referring to is in pcm_mulaw.c and is as
follows:-
static int snd_pcm_mulaw_hw_params(snd_pcm_t *pcm,
snd_pcm_hw_params_t
params) { snd_pcm_mulaw_t *mulaw = pcm->private_data; snd_pcm_format_t format; int err = snd_pcm_hw_params_slave(pcm, params,
snd_pcm_mulaw_hw_refine_cchange,
snd_pcm_mulaw_hw_refine_sprepare,
snd_pcm_mulaw_hw_refine_schange,
snd_pcm_generic_hw_params);
if (err < 0) return err; err =INTERNAL(snd_pcm_hw_params_get_format)(params, &format);
if (err < 0) return err; if (pcm->stream == SND_PCM_STREAM_PLAYBACK) { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx =snd_pcm_linear_get_index(format, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } else { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, mulaw->sformat); mulaw->func = snd_pcm_mulaw_decode; } } else { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, format); mulaw->func = snd_pcm_mulaw_decode; } else { mulaw->getput_idx = snd_pcm_linear_get_index(mulaw->sformat, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } } return 0; }
Thanks and regards, Mitul
-----Or
Hi,
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Tuesday, July 01, 2008 7:20 PM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
Hi Stan,
Thanks for all your help! I have some more questions though...
I downloaded the source code for alsa-lib-1.0.15 Based on
the code, if
the format is SND_PCM_FORMAT_MU_LAW, I am not sure why it does a get/put index to SND_PCM_FORMAT_S16 Also, if the stream is SND_PCM_STREAM_PLAYBACK, then I would think that it should
decode the
data. Why does it call snd_pcm_mulaw_decode function if the
format is
SND_PCM_FORMAT_MU_LAW and snd_pcm_mulaw_encode otherwise. I have an Intel HDA soundcard and according to the specs, it should
support PCM
ulaw format.
All ALSA documentation and examples I have come across use specific hw_params (like sample rate of 44100, SND_PCM_FORMAT_S16, 2 channel interleaved data). According to the documents, hw_params
refer to the
stream related info so that's the reason I tried to change
it to that
of mu-law (sampling rate of 8000 Hz, SND_PCM_FORMAT_MU_LAW
etc). Not
sure if that's the way to do it though. Based on the code it looks like the hardware just seems to support SND_PCM_FORMAT_S16. Any pointers to help me better understand the ALSA code would
be much appreciated.
Hi Misen,
First, a gentle remonstrance. You probably have noticed that I always put my responses after or mixed with your message. On public mailing lists this is considered good form, rather than posting your response at the top of the message. Why? So that anyone who steps into the interaction doesn't have to read the messages out of order and that future searchers have an easier time understanding the message. While top posting is the norm in communications between two or a few people because the context is familiar to all and it saves time not to have to look for the response, on a public mailing list that isn't necessarily true.
Now to the matter at hand. I had never heard of mu law so I looked it up. http://www.digitalpreservation.gov/formats/fdd/fdd000039.shtml ... Standard companding algorithm used in digital communications systems in North America and Japan (telephones, for the most part) to optimize the dynamic range of an analog signal (generally a voice) for digitizing, i.e., to compress 16 bit LPCM http://www.digitalpreservation.gov/formats/fdd/fdd000011.shtm l (Linear Pulse Code Modulated) data down to 8 bits of logarithmic data. See also Notes http://www.digitalpreservation.gov/formats/fdd/fdd000039.shtml#notes below. µ-Law is similar to the A-Law http://www.digitalpreservation.gov/formats/fdd/fdd000038.shtml algorithm used in Europe. ...
The code that you extracted below is designed to convert mu law from the compressed form back into the 16 bit signed form. I haven't checked the rest of the code myself, but it appears to assume that the sound device is incapable of internal conversion. If that is true, you shouldn't have to specify anything else to the library except mu law. It should take care of everything else. i.e. as soon as you specify mu law, it is known that the stream is 8 bit mono that has to be uncompressed to 16 bit mono. I presume that is why there is the error when you try to set the hardware parms with mu law. The library should probably be modified to use this new capability of sound device internal conversion for mu law if it is available on the sound device. Maybe it already does; as I said I haven't looked at the code, and I'm not really familiar with mu law.
I am making some changes to the alsa-lib code and I have built alsa-lib. But I don't think its really picking up my changes. How can I make sure that my application uses the modified library? I don't have to load any modules, do I? I am sorry if this is too basic a question but I couldn't find the info on a quick google search. Basically what I want to know is what are the steps to develop alsa-lib. This is what I did 1) Downloaded the source code 2) Configured the system using ./configure 3) Did a build using make 4) Did a "make install"
Am I missing something here?
Also regarding the original problem, when I run my program, the output of /proc/asound/card0/pcm0p/sub0/hw_params is
access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 48000 (48000/1) period_size: 32 buffer_size: 1024
This is clearly not what it should be since the data access should be RW_NONINTERLEAVED, format should be MU_LAW, there is only one channel and rate is 8000. Which would mean that alsa assumes a different set of parameters (for mu law)from what the data actually is. Am I right in thinking this?
Just got back after a long break and trying to pick up the threads again:-)
Again, any help will be much appreciated.
So, given my ignorance, my explanation and proposed solution might be completely wrong. :-) Perhaps a developer familiar with the coding of mu law will give a better explanation.
At this point, I really don't have more to offer for your problem. I would have to look at the code to decipher it in order to give an answer. You might as well do that yourself, as you will get a better understanding than I could give with an explanation.
The code that I am referring to is in pcm_mulaw.c and is as
follows:-
static int snd_pcm_mulaw_hw_params(snd_pcm_t *pcm,
snd_pcm_hw_params_t
params) { snd_pcm_mulaw_t *mulaw = pcm->private_data; snd_pcm_format_t format; int err = snd_pcm_hw_params_slave(pcm, params,
snd_pcm_mulaw_hw_refine_cchange,
snd_pcm_mulaw_hw_refine_sprepare,
snd_pcm_mulaw_hw_refine_schange,
snd_pcm_generic_hw_params);
if (err < 0) return err; err =INTERNAL(snd_pcm_hw_params_get_format)(params, &format);
if (err < 0) return err; if (pcm->stream == SND_PCM_STREAM_PLAYBACK) { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx =snd_pcm_linear_get_index(format, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } else { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, mulaw->sformat); mulaw->func = snd_pcm_mulaw_decode; } } else { if (mulaw->sformat == SND_PCM_FORMAT_MU_LAW) { mulaw->getput_idx = snd_pcm_linear_put_index(SND_PCM_FORMAT_S16, format); mulaw->func = snd_pcm_mulaw_decode; } else { mulaw->getput_idx = snd_pcm_linear_get_index(mulaw->sformat, SND_PCM_FORMAT_S16); mulaw->func = snd_pcm_mulaw_encode; } } return 0; }
Thanks and regards, Mitul
-----Or
Thanks, Mitul
Mitul Sen (misen) wrote:
I am making some changes to the alsa-lib code and I have built alsa-lib. But I don't think its really picking up my changes. How can I make sure that my application uses the
modified library? I don't have to load any modules, do I? I am sorry if this is too basic a question but I couldn't find the info on a quick google search. Basically what I want to know is what are the steps to develop alsa-lib. This is what I did
- Downloaded the source code
- Configured the system using ./configure
- Did a build using make
- Did a "make install"
Am I missing something here?
It looks correct. Do an ls -l /usr/lib/libasound*. The file there should have the same timestamp as the file in your build directory. If it doesn't, it didn't install. You could just copy it over.
Make sure there is no other copy in /lib.
Also regarding the original problem, when I run my program, the output of /proc/asound/card0/pcm0p/sub0/hw_params is
access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 48000 (48000/1) period_size: 32 buffer_size: 1024
This is clearly not what it should be since the data access should be RW_NONINTERLEAVED, format should be MU_LAW, there is only one channel and rate is 8000. Which would mean that alsa assumes a different set of parameters (for mu law)from what the data actually is. Am I right in thinking this?
Just got back after a long break and trying to pick up the threads again:-)
Again, any help will be much appreciated.
Unless your changes changed the mu-law code to use your card's mu-law decoder, they will still be the decoded values.
The best way to do this is to compile your program with debugging enabled ( -ggdb -O0) and the alsa library with debugging enabled. ./configure --help should give you the option. Then move only the library to /usr/lib and run your program as gdb --args yourprogram yourargs . You can see the info on how to run gdb in info gdb . With the debugger you can step through the program and the library to see where it is not working the way you expected. But you don't want to leave it like that as it is very inefficient.
If your code works for the mu-law, you could submit a patch. Of course, you would have to check for the functionality and branch to the old code if the card doesn't support mu-law decoding.
Hi,
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Wednesday, July 16, 2008 8:30 PM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
I am making some changes to the alsa-lib code and I have built alsa-lib. But I don't think its really picking up my
changes. How can
I make sure that my application uses the
modified library? I don't have to load any modules, do I? I am sorry if this is too basic a question but I couldn't find the info on a quick google search. Basically what I want to know is what are the steps to develop alsa-lib. This is what I did
- Downloaded the source code
- Configured the system using ./configure
- Did a build using make
- Did a "make install"
Am I missing something here?
It looks correct. Do an ls -l /usr/lib/libasound*. The file there should have the same timestamp as the file in your build directory. If it doesn't, it didn't install. You could just copy it over.
Make sure there is no other copy in /lib.
I have checked the timestamps for libasound.so and the soft links created in the /usr/lib directory after an install. They are the same as in my build directory.
Also regarding the original problem, when I run my program,
the output
of /proc/asound/card0/pcm0p/sub0/hw_params is
access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 48000 (48000/1) period_size: 32 buffer_size: 1024
This is clearly not what it should be since the data access
should be RW_NONINTERLEAVED, format should be MU_LAW, there is only one channel and rate is 8000. Which would mean that alsa assumes a different set of parameters (for mu law)from what the data actually is. Am I right in thinking this?
Just got back after a long break and trying to pick up the threads again:-)
Again, any help will be much appreciated.
Unless your changes changed the mu-law code to use your card's mu-law decoder, they will still be the decoded values.
The best way to do this is to compile your program with debugging enabled ( -ggdb -O0) and the alsa library with debugging enabled. ./configure --help should give you the option. Then move only the library to /usr/lib and run your program as gdb --args yourprogram yourargs . You can see the info on how to run gdb in info gdb . With the debugger you can step through the program and the library to see where it is not working the way you expected. But you don't want to leave it like that as it is very inefficient.
If your code works for the mu-law, you could submit a patch. Of course, you would have to check for the functionality and branch to the old code if the card doesn't support mu-law decoding.
I have tried using gdb both from the command line as you suggested and also from within eclipse. Even though I can step through the code and break properly, I think there is some mismatch between the source code and object code used by gdb. I say that because it sometimes steps through code in a way that makes no sense. For example, I see that a particular 'if' condition is satified and it goes into the 'if' clause and then again goes into the 'else' clause that is not expected. Is there any module that needs to be reloaded after building and installing the shared library? I have done a clean make at all times, checked timestamps, even rebooted the machine in case some driver related data needs to be reloaded at startup but none of this has helped.
Another thing that I notice is that when I use aplay to play the rtp data that I save to file (before writing to the sound device), and check the output of /proc/asound/card0/pcm0p/sub0/hw_params file, it is exactly the same as when I run my application. Using aplay does the playback properly even though hw_params still shows as
access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 48000 (48000/1)
Please note that I can play back the file using aplay, I only have the problem of bad audio when I try to write to the sound device in real-time. With this observation though I am not sure if the fact that the library seems to not use the card's decoder is really the problem. I am trying to look into the source code of aplay to see if I can spot any difference in the way the data is written to the buffer.
Meanwhile, any comments and help will be greatly appreciated as usual.
Thanks for your help.
Regards, Mitul
Mitul Sen (misen) wrote:
Hi,
I have tried using gdb both from the command line as you suggested and also from within eclipse. Even though I can step through the code and break properly, I think there is some mismatch between the source code and object code used by gdb. I say that because it sometimes steps through code in a way that makes no sense. For example, I see that a particular 'if' condition is satified and it goes into the 'if' clause and then again goes into the 'else' clause that is not expected. Is there any module that needs to be reloaded after building and installing the shared library? I have done a clean make at all times, checked timestamps, even rebooted the machine in case some driver related data needs to be reloaded at startup but none of this has helped.
I suspect you are debugging optimized code. The optimizer rearranges and deletes instructions. Did you specify -O0 so that no optimization occurs? The other gotcha in the alsa-lib code is that some of the functions are actually macros. They cannot be stepped through. When you hit them in the debugger it is disconcerting.
Another thing that I notice is that when I use aplay to play the rtp data that I save to file (before writing to the sound device), and check the output of /proc/asound/card0/pcm0p/sub0/hw_params file, it is exactly the same as when I run my application. Using aplay does the playback properly even though hw_params still shows as
access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 48000 (48000/1)
It is decoding it before it is playing it, it must be calling a routine somewhere to do that, or else it is built in.
Please note that I can play back the file using aplay, I only have the problem of bad audio when I try to write to the sound device in real-time. With this observation though I am not sure if the fact that the library seems to not use the card's decoder is really the problem. I am trying to look into the source code of aplay to see if I can spot any difference in the way the data is written to the buffer.
Good not to limit the possibilities you are examining.
Meanwhile, any comments and help will be greatly appreciated as usual.
Thanks for your help.
Regards, Mitul
Hi,
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Wednesday, July 23, 2008 11:26 AM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
Hi,
I have tried using gdb both from the command line as you
suggested and
also from within eclipse. Even though I can step through
the code and
break properly, I think there is some mismatch between the
source code
and object code used by gdb. I say that because it sometimes steps through code in a way that makes no sense. For example, I
see that a
particular 'if' condition is satified and it goes into the
'if' clause
and then again goes into the 'else' clause that is not expected. Is there any module that needs to be reloaded after building and installing the shared library? I have done a clean make at
all times,
checked timestamps, even rebooted the machine in case some driver related data needs to be reloaded at startup but none of
this has helped. I suspect you are debugging optimized code. The optimizer rearranges and deletes instructions. Did you specify -O0 so that no optimization occurs? The other gotcha in the alsa-lib code is that some of the functions are actually macros. They cannot be stepped through. When you hit them in the debugger it is disconcerting.
I did specify -O0 to disable optimizations.
Another thing that I notice is that when I use aplay to
play the rtp
data that I save to file (before writing to the sound device), and check the output of /proc/asound/card0/pcm0p/sub0/hw_params
file, it
is exactly the same as when I run my application. Using
aplay does the
playback properly even though hw_params still shows as
access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 48000 (48000/1)
It is decoding it before it is playing it, it must be calling a routine somewhere to do that, or else it is built in.
Please note that I can play back the file using aplay, I
only have the
problem of bad audio when I try to write to the sound device in real-time. With this observation though I am not sure if
the fact that
the library seems to not use the card's decoder is really
the problem.
I am trying to look into the source code of aplay to see if
I can spot
any difference in the way the data is written to the buffer.
Good not to limit the possibilities you are examining.
Meanwhile, any comments and help will be greatly
appreciated as usual.
Thanks for your help.
Regards, Mitul
Mitul Sen (misen) wrote:
Hi,
-----Original Message----- From: stan [mailto:ghjeold_i_mwee@cox.net] Sent: Wednesday, July 23, 2008 11:26 AM To: Mitul Sen (misen) Cc: alsa-devel@alsa-project.org Subject: Re: [alsa-devel] Setting format to SND_PCM_FORMAT_MU_LAW does not let me apply hardware parameters
Mitul Sen (misen) wrote:
Hi,
I have tried using gdb both from the command line as you
suggested and
also from within eclipse. Even though I can step through
the code and
break properly, I think there is some mismatch between the
source code
and object code used by gdb. I say that because it sometimes steps through code in a way that makes no sense. For example, I
see that a
particular 'if' condition is satified and it goes into the
'if' clause
and then again goes into the 'else' clause that is not expected. Is there any module that needs to be reloaded after building and installing the shared library? I have done a clean make at
all times,
checked timestamps, even rebooted the machine in case some driver related data needs to be reloaded at startup but none of
this has helped.
/sbin/ldconfig refreshes the links for libraries.
I suspect you are debugging optimized code. The optimizer rearranges and deletes instructions. Did you specify -O0 so that no optimization occurs? The other gotcha in the alsa-lib code is that some of the functions are actually macros. They cannot be stepped through. When you hit them in the debugger it is disconcerting.
I did specify -O0 to disable optimizations.
That's weird. Don't have an explanation except yours, that you aren't debugging the code you think you are.
participants (3)
-
Clemens Ladisch -
Mitul Sen (misen) -
stan