9:52 p.m.
Sound has never worked correctly for me since moving to a Linux 2.6
kernel. Maybe I just have an ultra-cheap sound card. In any case, the
existing linuxplay.c file doesn't do what my system wants. I wind up
with distorted sound, or no sound at all, and lots of warnings that too
few bytes were written to the sound device.
Here's a quick attempt by me to add support for ALSA. It isn't
complete. I'm posting it here to ask for comments and testers. This
patch "works", in the sense that I can successfully build with it, and
then I can play all of the sounds that come with Gnus. I don't yet know
what to do with the volume argument to play-sound and play-sound-file,
and the way I manage sound files is probably horrible in some way.
Furthermore, I'm not sure if the configure magic is right; does the
existence of the ALSA library imply that it should be used?
Put the attached file, alsaplay.c, in src/ and apply this patch to try
it out.
Index: configure.ac
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/configure.ac,v
retrieving revision 1.35
diff -d -u -r1.35 configure.ac
--- configure.ac 2006/02/27 16:29:06 1.35
+++ configure.ac 2006/03/20 21:44:37
＠＠ -5079,6 +5079,16 ＠＠
esac
fi
+ dnl Check for ALSA sound support
+ if test -z "$sound_found"; then
+ AC_CHECK_HEADER([alsa/pcm.h], [
+ sound_found=yes
+ AC_CHECK_LIB(asound, snd_pcm_open,
+ [with_native_sound_lib=-lasound
+ XE_ADD_OBJS(alsaplay.o)],
+ [sound_found=no])])
+ fi
+
dnl Check for Linux/BSD native sound (also on recent Cygwins)
if test -z "$sound_found"; then
for dir in "machine" "sys" "linux"; do
/* Play sound with the ALSA library
Copyright (C) 2006 Jerry James.
This file is part of XEmacs.
XEmacs is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.
XEmacs is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License
along with XEmacs; see the file COPYING. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA. */
/* Synched up with: Not in FSF. */
#include <config.h>
#include "lisp.h"
#include "sound.h"
#include "sysfile.h"
#include <alsa/input.h>
#include <alsa/output.h>
#include <alsa/conf.h>
#include <alsa/global.h>
#include <alsa/pcm.h>
#include <alsa/error.h>
#define WORD_LE(arr,start) (arr[start] + (arr[start + 1] << 8) + \
(arr[start + 2] << 16) + (arr[start + 3] << 24))
#define WORD_BE(arr,start) ((arr[start] << 24) + (arr[start + 1] << 16) + \
(arr[start + 2] << 8) + arr[start + 3])
/* This function was largely stolen from miscplay.c. */
static snd_pcm_format_t
analyze_format (Binbyte *format, int *speed, int *tracks)
{
/* Keep compatibility with Linux 68k, etc. by not relying on byte-sex */
if (!memcmp (format, "Creative Voice File\x1A\x1A\x00", 22) &&
(format[22] + (format[23] << 8)) ==
((0x1233 - format[24] - (format[25] << 8)) & 0xFFFF))
{
/* VOC */
*speed = 8000;
*tracks = 2;
return SND_PCM_FORMAT_U8;
}
else if (!memcmp (format, "RIFF", 4) && !memcmp (format + 8,
"WAVEfmt ", 8))
{
/* WAVE */
*speed = WORD_LE (format, 24);
*tracks = format[22];
return format[32] / format[22] == 1
? SND_PCM_FORMAT_U8
: SND_PCM_FORMAT_S16_LE;
}
else if (!memcmp (format, ".snd", 4))
{
/* Sun/NeXT Audio (big endian) */
if (WORD_BE (format, 4) < 24)
{
*speed = 8000;
*tracks = 1;
return SND_PCM_FORMAT_MU_LAW;
}
*speed = WORD_BE (format, 16);
*tracks = format[23];
if (!memcmp (format + 12, "\000\000\000", 3))
{
switch (format[15])
{
case 1:
case 17:
case 29:
return SND_PCM_FORMAT_MU_LAW;
case 2:
return SND_PCM_FORMAT_S8;
case 3:
return SND_PCM_FORMAT_S16_BE;
case 4:
return SND_PCM_FORMAT_S24_BE;
case 5:
return SND_PCM_FORMAT_S32_BE;
case 23:
case 24:
case 25:
case 26:
return SND_PCM_FORMAT_IMA_ADPCM;
case 27:
return SND_PCM_FORMAT_A_LAW;
default:
break;
}
}
return SND_PCM_FORMAT_UNKNOWN;
}
else if (!memcmp (format, ".sd", 4))
{
/* DEC Audio (little endian) */
if (WORD_LE (format, 4) < 24)
{
*speed = 8000;
*tracks = 1;
return SND_PCM_FORMAT_MU_LAW;
}
*speed = WORD_LE (format, 16);
*tracks = format[20];
if (!memcmp (format + 13, "\000\000\000", 3))
{
switch (format[12])
{
case 1:
case 17:
case 29:
return SND_PCM_FORMAT_MU_LAW;
case 2:
return SND_PCM_FORMAT_S8;
case 3:
return SND_PCM_FORMAT_S16_LE;
case 4:
return SND_PCM_FORMAT_S24_LE;
case 5:
return SND_PCM_FORMAT_S32_LE;
case 23:
case 24:
case 25:
case 26:
return SND_PCM_FORMAT_IMA_ADPCM;
case 27:
return SND_PCM_FORMAT_A_LAW;
default:
break;
}
}
return SND_PCM_FORMAT_UNKNOWN;
}
else
{
*speed = 8000;
*tracks = 1;
return SND_PCM_FORMAT_U8;
}
}
int
play_sound_data (Binbyte *data, int length, int volume)
{
snd_pcm_t *pcm_handle;
snd_pcm_hw_params_t *hwparams;
snd_pcm_format_t format;
int speed, tracks, err;
if ((err = snd_pcm_open (&pcm_handle, "default", SND_PCM_STREAM_PLAYBACK,
0)) < 0)
goto error_early;
snd_pcm_hw_params_alloca (&hwparams);
if ((err = snd_pcm_hw_params_any (pcm_handle, hwparams)) < 0)
goto error;
format = analyze_format (data, &speed, &tracks);
if ((err = snd_pcm_hw_params_set_access (pcm_handle, hwparams,
SND_PCM_ACCESS_RW_INTERLEAVED)) < 0)
goto error;
if ((err = snd_pcm_hw_params_set_format (pcm_handle, hwparams, format)) < 0)
goto error;
if ((err = snd_pcm_hw_params_set_rate (pcm_handle, hwparams, speed, 0)) < 0)
goto error;
if ((err = snd_pcm_hw_params_set_channels (pcm_handle, hwparams, tracks))
< 0)
goto error;
if ((err = snd_pcm_hw_params (pcm_handle, hwparams)) < 0)
goto error;
if ((err = snd_pcm_prepare (pcm_handle)) < 0)
goto error;
if ((err = snd_pcm_writei (pcm_handle, data, length)) < 0)
goto error;
snd_pcm_close (pcm_handle);
return 1;
error:
snd_pcm_close (pcm_handle);
error_early:
sound_perror (snd_strerror (err));
return 0;
}
void
play_sound_file (Extbyte *sound_file, int volume)
{
Binbyte *data;
int fd;
struct stat st;
fd = open (sound_file, O_RDONLY, 0);
if (fd < 0) {
sound_perror (sound_file);
return;
}
qxe_fstat (fd, &st);
data = xnew_array (Binbyte, st.st_size);
read (fd, data, st.st_size);
play_sound_data (data, st.st_size, volume);
xfree (data, Binbyte);
}
Regards,
--
Jerry James, Assistant Professor james(a)xemacs.org
Computer Science Department http://www.cs.usu.edu/~jerry/
Utah State University
5:50 a.m.
On 3/20/06, Jerry James <james(a)xemacs.org> wrote:
Sound has never worked correctly for me since moving to a Linux 2.6
kernel. Maybe I just have an ultra-cheap sound card. In any case, the
existing linuxplay.c file doesn't do what my system wants. I wind up
with distorted sound, or no sound at all, and lots of warnings that too
few bytes were written to the sound device.
Here's a quick attempt by me to add support for ALSA. It isn't
complete. I'm posting it here to ask for comments and testers. This
patch "works", in the sense that I can successfully build with it, and
then I can play all of the sounds that come with Gnus. I don't yet know
what to do with the volume argument to play-sound and play-sound-file,
and the way I manage sound files is probably horrible in some way.
Furthermore, I'm not sure if the configure magic is right; does the
existence of the ALSA library imply that it should be used?
yes, as long as the user can override it.
and your code looks ok to me.
Put the attached file, alsaplay.c, in src/ and apply this patch to try
it out.
Index: configure.ac
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/configure.ac,v
retrieving revision 1.35
diff -d -u -r1.35 configure.ac
--- configure.ac 2006/02/27 16:29:06 1.35
+++ configure.ac 2006/03/20 21:44:37
＠＠ -5079,6 +5079,16 ＠＠
esac
fi
+ dnl Check for ALSA sound support
+ if test -z "$sound_found"; then
+ AC_CHECK_HEADER([alsa/pcm.h], [
+ sound_found=yes
+ AC_CHECK_LIB(asound, snd_pcm_open,
+ [with_native_sound_lib=-lasound
+ XE_ADD_OBJS(alsaplay.o)],
+ [sound_found=no])])
+ fi
+
dnl Check for Linux/BSD native sound (also on recent Cygwins)
if test -z "$sound_found"; then
for dir in "machine" "sys" "linux"; do
/* Play sound with the ALSA library
Copyright (C) 2006 Jerry James.
This file is part of XEmacs.
XEmacs is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.
XEmacs is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License
along with XEmacs; see the file COPYING. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA. */
/* Synched up with: Not in FSF. */
#include <config.h>
#include "lisp.h"
#include "sound.h"
#include "sysfile.h"
#include <alsa/input.h>
#include <alsa/output.h>
#include <alsa/conf.h>
#include <alsa/global.h>
#include <alsa/pcm.h>
#include <alsa/error.h>
#define WORD_LE(arr,start) (arr[start] + (arr[start + 1] << 8) + \
(arr[start + 2] << 16) + (arr[start + 3] <<
24))
#define WORD_BE(arr,start) ((arr[start] << 24) + (arr[start + 1] << 16) +
\
(arr[start + 2] << 8) + arr[start + 3])
/* This function was largely stolen from miscplay.c. */
static snd_pcm_format_t
analyze_format (Binbyte *format, int *speed, int *tracks)
{
/* Keep compatibility with Linux 68k, etc. by not relying on
byte-sex */
if (!memcmp (format, "Creative Voice File\x1A\x1A\x00", 22) &&
(format[22] + (format[23] << 8)) ==
((0x1233 - format[24] - (format[25] << 8)) & 0xFFFF))
{
/* VOC */
*speed = 8000;
*tracks = 2;
return SND_PCM_FORMAT_U8;
}
else if (!memcmp (format, "RIFF", 4) && !memcmp (format + 8,
"WAVEfmt ",
8))
{
/* WAVE */
*speed = WORD_LE (format, 24);
*tracks = format[22];
return format[32] / format[22] == 1
? SND_PCM_FORMAT_U8
: SND_PCM_FORMAT_S16_LE;
}
else if (!memcmp (format, ".snd", 4))
{
/* Sun/NeXT Audio (big endian) */
if (WORD_BE (format, 4) < 24)
{
*speed = 8000;
*tracks = 1;
return SND_PCM_FORMAT_MU_LAW;
}
*speed = WORD_BE (format, 16);
*tracks = format[23];
if (!memcmp (format + 12, "\000\000\000", 3))
{
switch (format[15])
{
case 1:
case 17:
case 29:
return SND_PCM_FORMAT_MU_LAW;
case 2:
return SND_PCM_FORMAT_S8;
case 3:
return SND_PCM_FORMAT_S16_BE;
case 4:
return SND_PCM_FORMAT_S24_BE;
case 5:
return SND_PCM_FORMAT_S32_BE;
case 23:
case 24:
case 25:
case 26:
return SND_PCM_FORMAT_IMA_ADPCM;
case 27:
return SND_PCM_FORMAT_A_LAW;
default:
break;
}
}
return SND_PCM_FORMAT_UNKNOWN;
}
else if (!memcmp (format, ".sd", 4))
{
/* DEC Audio (little endian) */
if (WORD_LE (format, 4) < 24)
{
*speed = 8000;
*tracks = 1;
return SND_PCM_FORMAT_MU_LAW;
}
*speed = WORD_LE (format, 16);
*tracks = format[20];
if (!memcmp (format + 13, "\000\000\000", 3))
{
switch (format[12])
{
case 1:
case 17:
case 29:
return SND_PCM_FORMAT_MU_LAW;
case 2:
return SND_PCM_FORMAT_S8;
case 3:
return SND_PCM_FORMAT_S16_LE;
case 4:
return SND_PCM_FORMAT_S24_LE;
case 5:
return SND_PCM_FORMAT_S32_LE;
case 23:
case 24:
case 25:
case 26:
return SND_PCM_FORMAT_IMA_ADPCM;
case 27:
return SND_PCM_FORMAT_A_LAW;
default:
break;
}
}
return SND_PCM_FORMAT_UNKNOWN;
}
else
{
*speed = 8000;
*tracks = 1;
return SND_PCM_FORMAT_U8;
}
}
int
play_sound_data (Binbyte *data, int length, int volume)
{
snd_pcm_t *pcm_handle;
snd_pcm_hw_params_t *hwparams;
snd_pcm_format_t format;
int speed, tracks, err;
if ((err = snd_pcm_open (&pcm_handle, "default",
SND_PCM_STREAM_PLAYBACK,
0)) < 0)
goto error_early;
snd_pcm_hw_params_alloca (&hwparams);
if ((err = snd_pcm_hw_params_any (pcm_handle, hwparams)) < 0)
goto error;
format = analyze_format (data, &speed, &tracks);
if ((err = snd_pcm_hw_params_set_access (pcm_handle, hwparams,
SND_PCM_ACCESS_RW_INTERLEAVED))
< 0)
goto error;
if ((err = snd_pcm_hw_params_set_format (pcm_handle, hwparams, format))
< 0)
goto error;
if ((err = snd_pcm_hw_params_set_rate (pcm_handle, hwparams, speed, 0))
< 0)
goto error;
if ((err = snd_pcm_hw_params_set_channels (pcm_handle, hwparams,
tracks))
< 0)
goto error;
if ((err = snd_pcm_hw_params (pcm_handle, hwparams)) < 0)
goto error;
if ((err = snd_pcm_prepare (pcm_handle)) < 0)
goto error;
if ((err = snd_pcm_writei (pcm_handle, data, length)) < 0)
goto error;
snd_pcm_close (pcm_handle);
return 1;
error:
snd_pcm_close (pcm_handle);
error_early:
sound_perror (snd_strerror (err));
return 0;
}
void
play_sound_file (Extbyte *sound_file, int volume)
{
Binbyte *data;
int fd;
struct stat st;
fd = open (sound_file, O_RDONLY, 0);
if (fd < 0) {
sound_perror (sound_file);
return;
}
qxe_fstat (fd, &st);
data = xnew_array (Binbyte, st.st_size);
read (fd, data, st.st_size);
play_sound_data (data, st.st_size, volume);
xfree (data, Binbyte);
}
Regards,
--
Jerry James, Assistant Professor james(a)xemacs.org
Computer Science Department http://www.cs.usu.edu/~jerry/
Utah State University
5:17 a.m.
* Jerry James <james(a)xemacs.org> writes:
Here's a quick attempt by me to add support for ALSA.
Interesting how this thread has come up... we're currently doing
surgery on sound in SXEmacs.
,----[ From my `Installation'... ]
| Audio:
| Compiling in support for OSS (Open Sound System).
| Compiling in support for NAS (network audio system).
| Compiling in support for ESD (Enlightened Sound Daemon).
| Compiling in support for Polyp (a drop-in replacement for ESD).
| Compiling in support for ALSA (Advanced Linux Sound Architecture).
| Compiling in support for ao (generic audio output).
| Disabled audio drivers: aRts Jack
| Media Stream Handlers:
| Compiling in support for media streams handled internally.
| Compiling in support for media streams handled by sndfile.
| Compiling in support for media streams handled by ffmpeg.
| Compiling in support for media streams handled by sox.
| Compiling in support for media streams handled by mad.
| Disabled media drivers: xine gstreamer
`----
And as a teaser for you... (play-sound-file "file.mp3") works in
SXEmacs. Asynchronously, I might add. :-)
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| I am Dyslexic of Borg. |
| Fusistance is retile. Your arse will be laminated. |
|------------------------------------<steve(a)sxemacs.org>---|
5:59 p.m.
Steve Youngs <steve(a)sxemacs.org> wrote:
Interesting how this thread has come up... we're currently doing
surgery on sound in SXEmacs.
Cool. Are you making it part of the normal event processing system? I
was awfully disappointed when I went to look at the sound code and found
that it is entirely separate from normal events. I even thought briefly
about fixing that, but there are so many other things I want to do first
...
And as a teaser for you... (play-sound-file "file.mp3")
works in
SXEmacs. Asynchronously, I might add. :-)
:-) Well, great. Would you mind shoving me in the direction of
whatever documentation might tell me the number of tracks and the speed
of a given mp3?
Thanks for the reply, Steve.
--
Jerry James, Assistant Professor james(a)xemacs.org
Computer Science Department http://www.cs.usu.edu/~jerry/
Utah State University
3:30 a.m.
>>>> "Jerry" == Jerry James
<james(a)xemacs.org> writes:
Jerry> :-) Well, great. Would you mind shoving me in the
Jerry> direction of whatever documentation might tell me the
Jerry> number of tracks and the speed of a given mp3?
I'm really pressed for time (and need to go look at the archiving
breakage next, anyway) and don't have a URL offhand, but there was
copious documentation available from the MPEG site, tutorials, even.
It's patented stuff, so even the basic encoder algorithms were
published.
--
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
12:50 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Jerry James <james(a)xemacs.org> writes:
Steve Youngs <steve(a)sxemacs.org> wrote:
> Interesting how this thread has come up... we're currently doing
> surgery on sound in SXEmacs.
Cool. Are you making it part of the normal event processing system? I
was awfully disappointed when I went to look at the sound code and found
that it is entirely separate from normal events. I even thought briefly
Hm, why should normal events interact here? We use pthreads to trigger
some external libraries which do all the useful stuff for us. And at the
moment we are about to establish (elisp) sentinels which are called when
`events' (like stopping, seeking, or the like) happen to the stream. But
since such events are of external nature I cannot see how to wrap them
into lisp events. Can you? :)
about fixing that, but there are so many other things I want to do
first
...
Oh, what's the approach you have in mind?
> And as a teaser for you... (play-sound-file "file.mp3")
works in
> SXEmacs. Asynchronously, I might add. :-)
:-) Well, great. Would you mind shoving me in the direction of
whatever documentation might tell me the number of tracks and the speed
of a given mp3?
Hm, I cannot, sorry. Our implementation, as mentioned above, is mainly
driven by external libraries, so I can tell you how to do that with
ffmpeg/xine/gstreamer/libmad-APIs.
Oh, and with speed you mean the samplerate? Or the bitrate? You have to be
careful with the latter, since that may vary.
Sebastian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
iD8DBQFEKnSllMmhrILJOQ4RAsz6AJwIuu3IVLGc/yUYO1IIqERsIv9dpgCfQw5y
9+tGUKEJsTK/+hWa8YZElEE=
=nC0y
-----END PGP SIGNATURE-----
1:33 a.m.
On 3/29/06, Sebastian Freundt <hroptatyr(a)sxemacs.org> wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Jerry James <james(a)xemacs.org> writes:
> Steve Youngs <steve(a)sxemacs.org> wrote:
>> Interesting how this thread has come up... we're currently doing
>> surgery on sound in SXEmacs.
>
> Cool. Are you making it part of the normal event processing system? I
> was awfully disappointed when I went to look at the sound code and found
> that it is entirely separate from normal events. I even thought briefly
Hm, why should normal events interact here? We use pthreads to trigger
some external libraries which do all the useful stuff for us. And at the
moment we are about to establish (elisp) sentinels which are called when
`events' (like stopping, seeking, or the like) happen to the stream. But
since such events are of external nature I cannot see how to wrap them
into lisp events. Can you? :)
when you say "external" do you mean "need to be handled asynchronously in
lisp"? if so, then yes you can't put them in lisp events. but if no, then
you should use the same mechanism we currently use in order to convert ^G
under TTY's (which arrives as a signal) to a lisp event -- basically, we
select() on a pipe that the signal handler writes to.
ben
9:20 a.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
"Ben Wing" <benwing666(a)gmail.com> writes:
On 3/29/06, Sebastian Freundt <hroptatyr(a)sxemacs.org> wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Jerry James <james(a)xemacs.org> writes:
>
> > Steve Youngs <steve(a)sxemacs.org> wrote:
> >> Interesting how this thread has come up... we're currently doing
> >> surgery on sound in SXEmacs.
> >
> > Cool. Are you making it part of the normal event processing system? I
> > was awfully disappointed when I went to look at the sound code and found
> > that it is entirely separate from normal events. I even thought briefly
>
> Hm, why should normal events interact here? We use pthreads to trigger
> some external libraries which do all the useful stuff for us. And at the
> moment we are about to establish (elisp) sentinels which are called when
> `events' (like stopping, seeking, or the like) happen to the stream. But
> since such events are of external nature I cannot see how to wrap them
> into lisp events. Can you? :)
when you say "external" do you mean "need to be handled asynchronously in
lisp"? if so, then yes you can't put them in lisp events. but if no, then
The `need to be handled' is arguable, we support both asynchronous (==
pthreaded) and synchronous play. But basically asynchronous handling is
preferred.
you should use the same mechanism we currently use in order to
convert ^G
under TTY's (which arrives as a signal) to a lisp event -- basically, we
select() on a pipe that the signal handler writes to.
Hm, pipes, however, seem to be a good idea. What if the thread-trigger
(the function that pthread_create()s) installs a pipe which the thread
writes to and the event poller listens on that pipe.
On the other hand having a sentinel function, is it even necessary to use
the event mechanism? What's the benefit?
ben
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
iD8DBQFELOY5lMmhrILJOQ4RAqqFAJ4t304bdSlyRNf7tFNB/rJHldUQcgCcCYOl
s/45WhShry/jliqtacI+KFM=
=7ppW
-----END PGP SIGNATURE-----
6599
days inactive
6610
days old
7 comments
5 participants
participants (5)
-
Ben Wing -
Jerry James -
Sebastian Freundt -
Stephen J. Turnbull -
Steve Youngs