8 Jan
2007
8 Jan
'07
6:22 p.m.
Hello!
We're designing a new sound subsystem for allegro game programming
library, and we would like to take advantages of multiple hardware voice
capabilities.
On linux, ALSA is the only software that possibly could expose API for
such a capability, is that right? It is possible to open multiple PCM
outputs and play sound at the same time on all of them. The problem is
that we miss some basic mixing features, like panning a mono sound on
multichannel output for example. A similar feature is implemented as
'route' ALSA plug-in that can be specified in .asoundrc file, but it is
not flexible enough.
So the question is, does ALSA provide such functionality? If not, which
library could? Is it possible that linux OS doesn't provide any API to
take advantage of such hardware capabilities?
--
Milan Mimica
http://sparklet.sf.net
9 Jan
9 Jan
9:32 p.m.
Milan Mimica wrote:
Hello!
We're designing a new sound subsystem for allegro game programming
library, and we would like to take advantages of multiple hardware voice
capabilities.
On linux, ALSA is the only software that possibly could expose API for
such a capability, is that right? It is possible to open multiple PCM
outputs and play sound at the same time on all of them. The problem is
that we miss some basic mixing features, like panning a mono sound on
multichannel output for example. A similar feature is implemented as
'route' ALSA plug-in that can be specified in .asoundrc file, but it is
not flexible enough.
So the question is, does ALSA provide such functionality? If not, which
library could? Is it possible that linux OS doesn't provide any API to
take advantage of such hardware capabilities?
Surely this is a task that is up to the app to handle?
You would have to provide a function to move the mono signal across the
channel array accounting for volume. It would not be too difficult once
the base code for the channel handling was in place.
--
Patrick Shirkey - Boost Hardware Ltd.
Http://www.boosthardware.com
Http://lau.linuxaudio.org - The Linux Audio Users guide
========================================
"Anything your mind can see you can manifest physically, then it will
become reality" - Macka B
10 Jan
10 Jan
8:06 a.m.
Patrick Shirkey wrote:
Surely this is a task that is up to the app to handle?
That doesn't answer my question. What about unused fancy hardware? ALSA
obviously has the code ('route', 'dmix' and 'plug' plug-ins), it
just
doesn't provide the API (or I can't find it).
You would have to provide a function to move the mono
signal across the
channel array accounting for volume. It would not be too difficult once
the base code for the channel handling was in place.
We do have a software mixer already.
--
Milan Mimica
http://sparklet.sf.net
9:22 a.m.
On Mon, 2007-01-08 at 23:23 +0100, Milan Mimica wrote:
Hello!
We're designing a new sound subsystem for allegro game programming
library, and we would like to take advantages of multiple hardware voice
capabilities.
very few audio interfaces have this feature anymore. most h/w makers
seem to assume s/w mixing and no longer provide multiple PCM h/w
streams.
On linux, ALSA is the only software that possibly
could expose API for
such a capability, is that right?
its right and wrong. but mostly irrelevant.
It is possible to open multiple PCM
outputs and play sound at the same time on all of them.
if there *are* multiple h/w PCM voices, then ALSA provides "subdevices"
which get opened automatically each time you call snd_pcm_open() on a
given PCM device. however, as i mentioned, not much h/w provides this
any more. opening multiple PCM devices is entirely different and has its
own set of problems because the devices are generally not sample clock
synced.
The problem is
that we miss some basic mixing features, like panning a mono sound on
multichannel output for example. A similar feature is implemented as
'route' ALSA plug-in that can be specified in .asoundrc file, but it is
not flexible enough.
ALSA is not intended to provide any kind of live, real time signal
distribution API. this is a completely app-specific domain and it would
absurd for a generic HAL like ALSA to try to offer it. what you mean by
"panning" and the control you want is totally different than, say,
Ardour (http://ardour.org/) requires.
So the question is, does ALSA provide such
functionality? If not, which
library could? Is it possible that linux OS doesn't provide any API to
take advantage of such hardware capabilities?
in general, you should forget about the h/w capabilities of an audio
interface. for every user that has a device with some interesting
qualities, there will be 10 who do not.
welcome to winmodem for audio ...
--p
9:31 a.m.
Paul Davis wrote:
if there *are* multiple h/w PCM voices, then ALSA provides "subdevices"
which get opened automatically each time you call snd_pcm_open() on a
given PCM device.
Yes, that's what I meant.
ALSA is not intended to provide any kind of live, real
time signal
distribution API. this is a completely app-specific domain and it would
absurd for a generic HAL like ALSA to try to offer it. what you mean by
"panning" and the control you want is totally different than, say,
Ardour (http://ardour.org/) requires.
OK. This is the first time I got a straight "no". Well sort a :)
Thanks.
--
Milan Mimica
http://sparklet.sf.net
12:02 p.m.
Paul Davis wrote:
in general, you should forget about the h/w
capabilities of an audio
interface. for every user that has a device with some interesting
qualities, there will be 10 who do not.
welcome to winmodem for audio ...
Well, in this particular case, let's just see what the generic user out
there probably has. In quite a lot of cases it's some SoundBlaster
Audigy2 or X-Fi variant which both have somewhat extensive signal
processing capabilities (or that's what I've understood). Or you could
take ancient Aureal's Vortex2 as an example. Or also rather common
CMI8738 (not sure about this one). Current Linux support for these
devices is limited to PCM playback/recording and possibly bare MIDI.
Currently the thing we really are missing on Linux is HAL support for
something like OpenAL w/ EFX (or some DirectSound3D-for-Linux type
thing). Having this is somewhat crucial for game development. This is
also something OpenSL ES (http://www.khronos.org/opensles/) is trying to
address on embedded area.
- Jussi
12:17 p.m.
On 1/10/07, Jussi Laako <jussi.laako(a)pp.inet.fi> wrote:
Paul Davis wrote:
Well, not everyone. But given that the desktop station as to sell new/better
products, I would say this is still a marketing argument for any vendor.
But for the laptop (and most of the developers, who run laptops and don't
have money, and don't care too much), they probably have a very limited chip
that does not do hw mixing.
And what about HDA? Every laptop will soon support HDA. Does it means we
will have more hw mixing? As I said, it seems to be a marketing argument for
these technologies.
Currently the thing we really are missing on Linux is HAL support for
welcome to winmodem for audio ...
Well, in this particular case, let's just see what the generic user out
there probably has. In quite a lot of cases it's some SoundBlaster
Audigy2 or X-Fi variant which both have somewhat extensive signal something like OpenAL w/ EFX (or some
DirectSound3D-for-Linux type
thing). Having this is somewhat crucial for game development. This is
also something OpenSL ES (http://www.khronos.org/opensles/) is trying to
address on embedded area.
I feel the same, even if I would not criticize ALSA, which does "its-job".
But the initial question sounded more like a gentle request for current
status/state of art in linux audio. No offense here :)
Good luck to the next Allegro Audio API - that is an excellent initiative
(btw, you should consider a wrapper on top of PortAudio, i would suggest).
--
Marc-André Lureau, GSmartMix
12:23 p.m.
On Wed, 2007-01-10 at 18:02 +0200, Jussi Laako wrote:
Paul Davis wrote:
i suspect that the generic user at this point probably has an Intel HDA
device, the CMI or a SiS chipset.
i'm not a game developer and i can't comment on the usefulness of
DirectSound3D. it is just not used by the kinds of apps that i work on.
--p
in general, you should forget about the h/w
capabilities of an audio
interface. for every user that has a device with some interesting
qualities, there will be 10 who do not.
welcome to winmodem for audio ...
Well, in this particular case, let's just see what the generic user out
there probably has. In quite a lot of cases it's some SoundBlaster
Audigy2 or X-Fi variant which both have somewhat extensive signal
processing capabilities (or that's what I've understood). Or you could
take ancient Aureal's Vortex2 as an example. Or also rather common
CMI8738 (not sure about this one). Current Linux support for these
devices is limited to PCM playback/recording and possibly bare MIDI.
Currently the thing we really are missing on Linux is HAL support for
something like OpenAL w/ EFX (or some DirectSound3D-for-Linux type
thing). Having this is somewhat crucial for game development. This is
also something OpenSL ES (http://www.khronos.org/opensles/) is trying to
address on embedded area. 
13 Jan
13 Jan
12:20 a.m.
Milan Mimica wrote:
Hello!
We're designing a new sound subsystem for allegro game programming
library, and we would like to take advantages of multiple hardware
voice capabilities.
On linux, ALSA is the only software that possibly could expose API for
such a capability, is that right? It is possible to open multiple PCM
outputs and play sound at the same time on all of them. The problem is
that we miss some basic mixing features, like panning a mono sound on
multichannel output for example. A similar feature is implemented as
'route' ALSA plug-in that can be specified in .asoundrc file, but it
is not flexible enough.
So the question is, does ALSA provide such functionality? If not,
which library could? Is it possible that linux OS doesn't provide any
API to take advantage of such hardware capabilities?
For games programming, I suggest you use the OpenAL API.
OpenAL will use the hardware capabilities of the sound card, if present,
leaving the game to not worry if the work is done in hardware or software.
One extra suggestion I would make is to make sure the game uses 48kHz
audio, and not 44.1kHz audio, as not many sound cards support 44.1kHz in
hardware, resulting is CPU intensive resampling.
If I missunderstood what you are doing, and you instead wish to
implement your own OpenAL type API, then you can ask alsa-lib to tell
you how many channels the hardware can mix together.
James
7:58 a.m.
James Courtier-Dutton wrote:
If I missunderstood what you are doing, and you instead wish to
implement your own OpenAL type API, then you can ask alsa-lib to tell
you how many channels the hardware can mix together.
Allegro provides a unified API to various sound interfaces on various
platforms. The programmer (or the user) can choose from several output
drivers: ALSA, OSS, arts, JACK, OpenAL, DicestSound, ... But none of
those is required though!
We already have a OpenAL wrapper implemented and OpenAL provides all we
need. Now we would like to the same thing with ALSA, because ALSA is
more common on linux installations than OpenAL. Of course, we can do it
with our software mixer + ALSA, but we where thinking of leaving mixing
to the hardware if possible.
For games programming, I suggest you use the OpenAL
API.
OpenAL will use the hardware capabilities of the sound card, if present,
leaving the game to not worry if the work is done in hardware or
software.
OpenAL can use hardware capabilities of the sound card? Hm, can it do
mixing in hardware too? Maybe I just have to check how do they implement
it in linux.
One extra suggestion I would make is to make sure the
game uses 48kHz
audio, and not 44.1kHz audio, as not many sound cards support 44.1kHz in
hardware, resulting is CPU intensive resampling.
It's up to the programmer (library user) to decide which sample rate to use.
--
Milan Mimica
http://sparklet.sf.net
7:16 p.m.
Milan Mimica wrote:
If "mixing in hardware" is one of the "hardware capabilities", then
yes.
Not all sound cards can do hardware mixing.
If these are ALSA specific questions, then you might do better asking on
the alsa-user mailing list instead of this one.
ALSA presents a very simple picture regarding hardware mixing.
If the device has more than one sub PCM device, it can do hardware mixing.
/proc/asound/card0/pcm0p
Then "ls" will display the sub devices.
There is an api call to retrieve it as a simple integer, but I don't
remember the call off hand.
I.e. return 1 for not hardware mixing, and >1 for hardware mixing, with
the returned integer indicating how many channels can be mixed together
in hardware.
To do the hardware mixing, one simply opens the alsa device name, e.g.
"plug:front" multiple times, and the hardware will mix them together.
Do not use the alsa device names starting with hw: . These are for
use internally by alsa-lib, and should not be used by applications.
Is that what you need?
James
For games programming, I suggest you use the
OpenAL API.
OpenAL will use the hardware capabilities of the sound card, if
present,
leaving the game to not worry if the work is done
in hardware or
software.
OpenAL can use hardware capabilities of the sound card? Hm, can it do
mixing in hardware too? Maybe I just have to check how do they
implement it in linux.
8:08 p.m.
James Courtier-Dutton wrote:
To do the hardware mixing, one simply opens the alsa device name, e.g.
"plug:front" multiple times, and the hardware will mix them together.
I know so far. But I would like to get a bit more from ALSA. For
example, open a 'slave' 2-ch device, and open some mono subdevices that
will be 'routed' to that slave, according to 'route table'. ...If
you're
familiar with ALSA terminology.
Something similar can be done using .asoundrc configuration file, so I
assume that the code exists. Furthermore, there is ALSA plug-in API that
lets me create a 'routed' subdevice. The only problem is that, once set,
the 'route table' cannot be changed.
As it has been said in this thread, this is probably out of ALSA's domain.
--
Milan Mimica
http://sparklet.sf.net
14 Jan
14 Jan
8:25 a.m.
New subject: [linux-audio-dev] Saving gtk slider settings
Hi,
I'm looking at the save feature for jackeq and haven't found an easy to
grok example of saving the standard gtk slider settings.
Can someone point me to a specific function that saves the settings for
standard gtk2 sliders.
Cheers.
--
Patrick Shirkey - Boost Hardware Ltd.
Http://www.boosthardware.com
Http://lau.linuxaudio.org - The Linux Audio Users guide
========================================
"Anything your mind can see you can manifest physically, then it will
become reality" - Macka B
6319
days inactive
6325
days old
linux-audio-dev@lists.linuxaudio.org
12 comments
6 participants
participants (6)
-
James Courtier-Dutton -
Jussi Laako -
Marc-André Lureau -
Milan Mimica -
Patrick Shirkey -
Paul Davis