[alsa-devel] Sound streaming application: which device to use?
Hi all,
I'm completely new to ALSA, so forgive me if these are newby questions.
I'm working on an application which receives sound chunks via an IP connection, and it should use the local sound card to make this sound audible. Up to now it used OSS, but there were some problems with that, so I'm busy with rewriting the application to ALSA. The sound is 16-bit stereo, sampled at 8000 Hz, and it comes in chunks of 1024 frames. I managed to get things more or less working, but there are still some things unclear to me.
1. If I open the soundcard with device "default" or device "plughw" I get good sound, but the device "default" only accepts 1020 frames of every 1024 that I offer it (as can be seen from the return value of snd_pcm_writei() ). The device "plughw" nicely accepts all 1024 frames. Why is this? In both cases I have set the sample rate to 8000 via a call to snd_pcm_hw_params_set_rate_near() and the buffer size to 2048 frames via snd_pcm_hw_params_set_buffer_size_near().
2. If I use device "default", there seems to be no obligation to set the buffer size nor the period size, I still get good sound at the output. If I use "plughw", I have to set either the buffer size (2048 frames) or the period size (1024 frames), otherwise I get distorted sound. If I set both (the buffer size being 2048 and the period size being 1024) I also get distorted sound. Why?
3. If I set the buffer size twice the period size (as above), and ask the sizes back, ALSA nicely reports the buffer size as being twice the period size. But if I ask the buffer and period TIMES, they are reported as being the same: 128000 us! Huh????
4. The return value of snd_pcm_hw_params_set_rate_near() is sometimes 8000, sometimes 7999, from the very same program. I would expect ALSA to always return the same value, be it what I request or not, but at least being deterministic. How can this be?
5. There is another application running on the same target system, which will sometimes produce audible alarms. However, if my application is running that other application cannot generate the alarms, it produces a "resource busy" error. It is clear to me that my application is using the sound card exclusively, but how can I get it to allow other applications access to it too? I tried using the device "dmix", but then my sound is again very distorted.
Hope that somebody can help me out. Thanks in advance.
Lex
'Twas brillig, and Lex Wassenberg at 04/06/09 12:28 did gyre and gimble:
I'm working on an application which receives sound chunks via an IP connection, and it should use the local sound card to make this sound audible. Up to now it used OSS, but there were some problems with that, so I'm busy with rewriting the application to ALSA. The sound is 16-bit stereo, sampled at 8000 Hz, and it comes in chunks of 1024 frames. I managed to get things more or less working, but there are still some things unclear to me.
Depending on the use case of your application, have you considered writing a pulseaudio module instead.
Pulseaudio already integrates with wireless and network systems for audio, notably bluetooth, Apple RAOP/Airtunes, and it's uPnP media stuff, RTP streams and it's own network protocol.
It would be fairly easy to hook up a remote device like this and have it "play" the sound via pulseuaudio.
You'll be needing some kind of daemon/background process anyway to listen on the network so this may be a better construct for you to work with rather than having to worry to start your "listener" app.
There may be many reasons why this wouldn't work for you but figured it was worth mentioning.
Col
On Thu, Jun 4, 2009 at 1:28 PM, Lex Wassenberglex.wassenberg@gmail.com wrote:
Hi all,
I'm completely new to ALSA, so forgive me if these are newby questions.
I'm working on an application which receives sound chunks via an IP connection, and it should use the local sound card to make this sound audible. Up to now it used OSS, but there were some problems with that, so I'm busy with rewriting the application to ALSA. The sound is 16-bit stereo, sampled at 8000 Hz, and it comes in chunks of 1024 frames. I managed to get things more or less working, but there are still some things unclear to me.
- If I open the soundcard with device "default" or device "plughw" I
get good sound, but the device "default" only accepts 1020 frames of every 1024 that I offer it (as can be seen from the return value of snd_pcm_writei() ). The device "plughw" nicely accepts all 1024 frames. Why is this? In both cases I have set the sample rate to 8000 via a call to snd_pcm_hw_params_set_rate_near() and the buffer size to 2048 frames via snd_pcm_hw_params_set_buffer_size_near().
- If I use device "default", there seems to be no obligation to set
the buffer size nor the period size, I still get good sound at the output. If I use "plughw", I have to set either the buffer size (2048 frames) or the period size (1024 frames), otherwise I get distorted sound. If I set both (the buffer size being 2048 and the period size being 1024) I also get distorted sound. Why?
- If I set the buffer size twice the period size (as above), and ask
the sizes back, ALSA nicely reports the buffer size as being twice the period size. But if I ask the buffer and period TIMES, they are reported as being the same: 128000 us! Huh????
- The return value of snd_pcm_hw_params_set_rate_near() is sometimes
8000, sometimes 7999, from the very same program. I would expect ALSA to always return the same value, be it what I request or not, but at least being deterministic. How can this be?
- There is another application running on the same target system,
which will sometimes produce audible alarms. However, if my application is running that other application cannot generate the alarms, it produces a "resource busy" error. It is clear to me that my application is using the sound card exclusively, but how can I get it to allow other applications access to it too? I tried using the device "dmix", but then my sound is again very distorted.
Hope that somebody can help me out. Thanks in advance.
Lex
Is there nobody that can help me out on the questions above? I figured the answer to question 4 (passing a pointer to an uninitialized integer; I thought the integer was only used for reporting back, which is not true), but all other questions still stand. Nobody??
Lex
On Tue, Jun 9, 2009 at 10:01 AM, Lex Wassenberglex.wassenberg@gmail.com wrote:
Is there nobody that can help me out on the questions above? I figured the answer to question 4 (passing a pointer to an uninitialized integer; I thought the integer was only used for reporting back, which is not true), but all other questions still stand. Nobody??
You don't provide enough info. What soundcard? What distro? What is the output of "aplay -v file.wav" when file.wav is the same audio format your app uses?
Lee
On Wed, Jun 10, 2009 at 4:42 AM, Lee Revellrlrevell@joe-job.com wrote:
On Tue, Jun 9, 2009 at 10:01 AM, Lex Wassenberglex.wassenberg@gmail.com wrote:
Is there nobody that can help me out on the questions above? I figured the answer to question 4 (passing a pointer to an uninitialized integer; I thought the integer was only used for reporting back, which is not true), but all other questions still stand. Nobody??
You don't provide enough info. What soundcard? What distro? What is the output of "aplay -v file.wav" when file.wav is the same audio format your app uses?
Lee
The distro is OpenSuse 10.3 , the soundcard doesn't (or shouldn't) matter: the application program knows nothing about the soundcard and it doesn't want to either. It should just use that sound card that is present on the machine where it runs. I thought that it would be ALSA that takes care of the particularites of the underlying hardware. That's why I programmed the device to be "default" or "plughw".
For the requested output of the aplay program I would first have to find a wave file with the specific format.
Lex
Lex Wassenberg wrote:
On Wed, Jun 10, 2009 at 4:42 AM, Lee Revellrlrevell@joe-job.com wrote:
On Tue, Jun 9, 2009 at 10:01 AM, Lex Wassenberglex.wassenberg@gmail.com wrote:
Is there nobody that can help me out on the questions above? I figured the answer to question 4 (passing a pointer to an uninitialized integer; I thought the integer was only used for reporting back, which is not true), but all other questions still stand. Nobody??
You don't provide enough info. What soundcard? What distro? What is the output of "aplay -v file.wav" when file.wav is the same audio format your app uses?
Lee
The distro is OpenSuse 10.3 , the soundcard doesn't (or shouldn't) matter: the application program knows nothing about the soundcard and it doesn't want to either. It should just use that sound card that is present on the machine where it runs. I thought that it would be ALSA that takes care of the particularites of the underlying hardware. That's why I programmed the device to be "default" or "plughw".
For the requested output of the aplay program I would first have to find a wave file with the specific format.
Just use sox, it will convert to any format you need. Details man sox, e.g.
sox your.mp3 -r 44100 -b 16 new-44k-16.wav
Pavel.
participants (4)
-
Colin Guthrie -
Lee Revell -
Lex Wassenberg -
Pavel Hofman