Surround sound and volume control question

Surround sound and volume control question

[Daniel Jacobowitz]

How is volume control supposed to work for surround sound codecs?  Is there
typically a register which affects every channel, or do you have to adjust
them separately?  I've been staring at the datasheets and prodding the
registers for mine (ICE1724, on a Shuttle SN25P).  There's three pieces:
the vt1720, the vt1617a, and a wm8728.  The vt1617a has ac97 volume control
registers, but the "master" volume doesn't affect the center/lfe or rear
surround channels.  I am guessing that the wm8728 is just for the last two
channels (I've got jacks for 7.1 but the vt1617a only does 5.1).

I cobbled together patches that add a stereo mute switch to the softvol
plugin, and set that in front of the card.  Then I can actually change
volume or mute the whole set.  But for various reasons this is
unsatisfactory, e.g. because it's not called "Master Playback", and because
it doesn't show up as the first mixer element - every application needs a
bit of configuration to find the right volume control.  And one of the
applications I wanted to configure, gnome-settings-daemon, is basically
unconfigurable.  It just grabs whatever gstreamer sees at the front of the
list.

(The mute patches available on request, btw, one alsa-lib patch and one
kernel driver bugfix for user-created boolean controls.)

Before I go any further with this I want to know how it works elsewhere.
I would think that having a single mixer control that affected all the
surround speakers would be a pretty standard thing to want!  Do other chips
have a single register for it?  Or does this one, and I'm just not seeing
it?

If I have no other way to get this to work than with the softvol plugin,
I think this is the way to go:
  - Rename the "Master" volume control/switch to "Front".
  - In userspace, have softvol create a control/switch named "Master".

Potential problems:
  - With a library older than kernel driver I think things would start
    picking up the Headphone channel as the master volume control if
    I made this change, yuck.
  - The user-created controls don't exist right away, so saving and
    restoring volume might not work.

All comments very welcome - I'm out of my depth here.  There's a bit of
related discussion in bug 1407.

-- 
Daniel Jacobowitz
CodeSourcery

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Takashi Iwai]

At Sun, 8 Oct 2006 18:27:26 -0400,
Daniel Jacobowitz wrote:
> 
> How is volume control supposed to work for surround sound codecs?  Is there
> typically a register which affects every channel, or do you have to adjust
> them separately?  I've been staring at the datasheets and prodding the
> registers for mine (ICE1724, on a Shuttle SN25P).  There's three pieces:
> the vt1720, the vt1617a, and a wm8728.  The vt1617a has ac97 volume control
> registers, but the "master" volume doesn't affect the center/lfe or rear
> surround channels.  I am guessing that the wm8728 is just for the last two
> channels (I've got jacks for 7.1 but the vt1617a only does 5.1).

That's the standard behavior of AC97 codec.  The "master" doesn't
influence on surrounds.  Some ac97 codec chips have an extended mode
that behaves differently, though.

> I cobbled together patches that add a stereo mute switch to the softvol
> plugin, and set that in front of the card.  Then I can actually change
> volume or mute the whole set.  But for various reasons this is
> unsatisfactory, e.g. because it's not called "Master Playback", and because
> it doesn't show up as the first mixer element - every application needs a
> bit of configuration to find the right volume control.  And one of the
> applications I wanted to configure, gnome-settings-daemon, is basically
> unconfigurable.  It just grabs whatever gstreamer sees at the front of the
> list.
> 
> (The mute patches available on request, btw, one alsa-lib patch and one
> kernel driver bugfix for user-created boolean controls.)
> 
> Before I go any further with this I want to know how it works elsewhere.
> I would think that having a single mixer control that affected all the
> surround speakers would be a pretty standard thing to want!  Do other chips
> have a single register for it?  Or does this one, and I'm just not seeing
> it?
> 
> If I have no other way to get this to work than with the softvol plugin,
> I think this is the way to go:
>   - Rename the "Master" volume control/switch to "Front".
>   - In userspace, have softvol create a control/switch named "Master".

Maybe it's an easiest way...

> Potential problems:
>   - With a library older than kernel driver I think things would start
>     picking up the Headphone channel as the master volume control if
>     I made this change, yuck.
>   - The user-created controls don't exist right away, so saving and
>     restoring volume might not work.

alsactl save & restore should work for softvol.
The softvol mixer element doesn't exist only at the very fresh state.


Takashi

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Daniel Jacobowitz]

On Mon, Oct 09, 2006 at 03:37:49PM +0200, Takashi Iwai wrote:
> At Sun, 8 Oct 2006 18:27:26 -0400,
> Daniel Jacobowitz wrote:
> > 
> > How is volume control supposed to work for surround sound codecs?  Is there
> > typically a register which affects every channel, or do you have to adjust
> > them separately?  I've been staring at the datasheets and prodding the
> > registers for mine (ICE1724, on a Shuttle SN25P).  There's three pieces:
> > the vt1720, the vt1617a, and a wm8728.  The vt1617a has ac97 volume control
> > registers, but the "master" volume doesn't affect the center/lfe or rear
> > surround channels.  I am guessing that the wm8728 is just for the last two
> > channels (I've got jacks for 7.1 but the vt1617a only does 5.1).
> 
> That's the standard behavior of AC97 codec.  The "master" doesn't
> influence on surrounds.  Some ac97 codec chips have an extended mode
> that behaves differently, though.

I assume this one doesn't.  But could you tell me one supported card
which does, so that I can compare?  Thanks.

> alsactl save & restore should work for softvol.
> The softvol mixer element doesn't exist only at the very fresh state.

I see, I just tried it and alsactl will create a user control just
fine.  That's great.

I will give the softvol solution a try, and post the patches if it
works.

-- 
Daniel Jacobowitz
CodeSourcery

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Takashi Iwai]

At Mon, 9 Oct 2006 09:45:39 -0400,
Daniel Jacobowitz wrote:
> 
> On Mon, Oct 09, 2006 at 03:37:49PM +0200, Takashi Iwai wrote:
> > At Sun, 8 Oct 2006 18:27:26 -0400,
> > Daniel Jacobowitz wrote:
> > > 
> > > How is volume control supposed to work for surround sound codecs?  Is there
> > > typically a register which affects every channel, or do you have to adjust
> > > them separately?  I've been staring at the datasheets and prodding the
> > > registers for mine (ICE1724, on a Shuttle SN25P).  There's three pieces:
> > > the vt1720, the vt1617a, and a wm8728.  The vt1617a has ac97 volume control
> > > registers, but the "master" volume doesn't affect the center/lfe or rear
> > > surround channels.  I am guessing that the wm8728 is just for the last two
> > > channels (I've got jacks for 7.1 but the vt1617a only does 5.1).
> > 
> > That's the standard behavior of AC97 codec.  The "master" doesn't
> > influence on surrounds.  Some ac97 codec chips have an extended mode
> > that behaves differently, though.
> 
> I assume this one doesn't.  But could you tell me one supported card
> which does, so that I can compare?  Thanks.

AFAIK, AD198x codecs have an non-compatible mode.  It's controlled via
register 0x76 bit 14.

> > alsactl save & restore should work for softvol.
> > The softvol mixer element doesn't exist only at the very fresh state.
> 
> I see, I just tried it and alsactl will create a user control just
> fine.  That's great.
> 
> I will give the softvol solution a try, and post the patches if it
> works.

That's nice.  Thanks.


Takashi

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Lee Revell]

On Mon, 2006-10-09 at 09:45 -0400, Daniel Jacobowitz wrote:
> > alsactl save & restore should work for softvol.
> > The softvol mixer element doesn't exist only at the very fresh state.
> 
> I see, I just tried it and alsactl will create a user control just
> fine.  That's great.
> 
> I will give the softvol solution a try, and post the patches if it
> works.
> 

Isn't it inefficient to force all audio to go through a software volume
control just to have a master volume?  Wouldn't it be better to make a
ganged control in the driver (until the alsa-lib abstract mixer layer is
ready)?

Lee


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Daniel Jacobowitz]

On Mon, Oct 09, 2006 at 09:55:23AM -0400, Lee Revell wrote:
> On Mon, 2006-10-09 at 09:45 -0400, Daniel Jacobowitz wrote:
> > > alsactl save & restore should work for softvol.
> > > The softvol mixer element doesn't exist only at the very fresh state.
> > 
> > I see, I just tried it and alsactl will create a user control just
> > fine.  That's great.
> > 
> > I will give the softvol solution a try, and post the patches if it
> > works.
> > 
> 
> Isn't it inefficient to force all audio to go through a software volume
> control just to have a master volume?  Wouldn't it be better to make a
> ganged control in the driver (until the alsa-lib abstract mixer layer is
> ready)?

I did consider doing it this way over the weekend.  However, being able
to adjust the relative volumes of the different channels still seems
useful to me.  If you have a single unified control, then you mess up
trying to move the others.

What do you think?  Should I just slave them all together and remove
the other volume controls?  It's probably easier; it just seems somehow
lacking.

-- 
Daniel Jacobowitz
CodeSourcery

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Lee Revell]

On Mon, 2006-10-09 at 10:38 -0400, Daniel Jacobowitz wrote:
> I did consider doing it this way over the weekend.  However, being
> able to adjust the relative volumes of the different channels still
> seems useful to me.  If you have a single unified control, then you
> mess up trying to move the others.
> 
> What do you think?  Should I just slave them all together and remove
> the other volume controls?  It's probably easier; it just seems
> somehow lacking.
> 

How do other OSes do it?  Do they also force everything through a
software volume control?

Is a ganged master control fundamentally incompatible with individual
channel volumes?  If so, doesn't it mean that this problem can never be
solved by an abstract mixer layer?

Lee


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Takashi Iwai]

At Mon, 09 Oct 2006 10:42:56 -0400,
Lee Revell wrote:
> 
> On Mon, 2006-10-09 at 10:38 -0400, Daniel Jacobowitz wrote:
> > I did consider doing it this way over the weekend.  However, being
> > able to adjust the relative volumes of the different channels still
> > seems useful to me.  If you have a single unified control, then you
> > mess up trying to move the others.
> > 
> > What do you think?  Should I just slave them all together and remove
> > the other volume controls?  It's probably easier; it just seems
> > somehow lacking.
> > 
> 
> How do other OSes do it?  Do they also force everything through a
> software volume control?

I've not seen any codes of Windows driver, but most likely they do
software volume and effects control.

> Is a ganged master control fundamentally incompatible with individual
> channel volumes?  If so, doesn't it mean that this problem can never be
> solved by an abstract mixer layer?

It's possible to create a virtual volume element that represents the
base attenuation level.  Suppose "Master" to be a virtual volume (a
mono), then the real volume values for front, rear and CLFE become
"front + master", "rear + master" and "CLFE + master".  The values
have to be clipped to max 31 (or 63).

The problem is that the mixer values are no longer corresponding to
the register values as 1:1.  Thus the current code for get/put
callbacks can't be used as it is.  The values have to be cached and
handled internally independent from the register values...


Takashi

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Daniel Jacobowitz]

On Mon, Oct 09, 2006 at 04:50:49PM +0200, Takashi Iwai wrote:
> > Is a ganged master control fundamentally incompatible with individual
> > channel volumes?  If so, doesn't it mean that this problem can never be
> > solved by an abstract mixer layer?
> 
> It's possible to create a virtual volume element that represents the
> base attenuation level.  Suppose "Master" to be a virtual volume (a
> mono), then the real volume values for front, rear and CLFE become
> "front + master", "rear + master" and "CLFE + master".  The values
> have to be clipped to max 31 (or 63).
> 
> The problem is that the mixer values are no longer corresponding to
> the register values as 1:1.  Thus the current code for get/put
> callbacks can't be used as it is.  The values have to be cached and
> handled internally independent from the register values...

Do you think this is a better choice than using softvol?  If so, I
will work on the implementation later this week.

-- 
Daniel Jacobowitz
CodeSourcery

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Takashi Iwai]

At Mon, 9 Oct 2006 13:04:51 -0400,
Daniel Jacobowitz wrote:
> 
> On Mon, Oct 09, 2006 at 04:50:49PM +0200, Takashi Iwai wrote:
> > > Is a ganged master control fundamentally incompatible with individual
> > > channel volumes?  If so, doesn't it mean that this problem can never be
> > > solved by an abstract mixer layer?
> > 
> > It's possible to create a virtual volume element that represents the
> > base attenuation level.  Suppose "Master" to be a virtual volume (a
> > mono), then the real volume values for front, rear and CLFE become
> > "front + master", "rear + master" and "CLFE + master".  The values
> > have to be clipped to max 31 (or 63).
> > 
> > The problem is that the mixer values are no longer corresponding to
> > the register values as 1:1.  Thus the current code for get/put
> > callbacks can't be used as it is.  The values have to be cached and
> > handled internally independent from the register values...
> 
> Do you think this is a better choice than using softvol?  If so, I
> will work on the implementation later this week.

Depends.  It's definitely ligher for hardware at least, but would
involve more changes in the driver side.  OTOH, softvol can provide
finer volume controls with higher resolution.

I thought of implementation of virtual mixer element in alsa-lib, but
it'd be also a bit messy.  The implementation in the driver side could
be even easier.


Takashi

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Daniel Jacobowitz]

On Mon, Oct 09, 2006 at 10:42:56AM -0400, Lee Revell wrote:
> On Mon, 2006-10-09 at 10:38 -0400, Daniel Jacobowitz wrote:
> > I did consider doing it this way over the weekend.  However, being
> > able to adjust the relative volumes of the different channels still
> > seems useful to me.  If you have a single unified control, then you
> > mess up trying to move the others.
> > 
> > What do you think?  Should I just slave them all together and remove
> > the other volume controls?  It's probably easier; it just seems
> > somehow lacking.
> > 
> 
> How do other OSes do it?  Do they also force everything through a
> software volume control?

Unfortunately, I could find a program to dump the ice1724's IO space in
Windows, but I couldn't find anything to dump the vt1617a AC97
registers (which require writes to select the registers IIRC).  If
you've got a recommendation, I can boot this machine into Windows and
see what happens to the vt1617a's registers when the Windows driver
adjusts the volume.

> Is a ganged master control fundamentally incompatible with individual
> channel volumes?  If so, doesn't it mean that this problem can never be
> solved by an abstract mixer layer?

I don't know.  I guess you could preserve individual channel volumes by
saving them separately (outside the registers), and using a volume of
"master control plus channel control".  But I don't know how the
resulting quality would be or how tricky it would be to implement.
Right now we never cache the volumes corresponding to hardware
registers.

-- 
Daniel Jacobowitz
CodeSourcery

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Surround sound and volume control question

[Anders Johansson]

Hi,

Just wanted to give you some input from a user of the more advanced
features of alsa volume controls. I am running an acoustic lab at the
University of Western Australia. I have previously helped out in alsa
to reverse engineer a sound card and fix some minor bugs. We use
alsa/jack to do sound recordings and to implement realtime audio
processing algorithms where the absolute relationship between the
volume level and the voltage level at the sound card output must be
known. Currently I have had to reverse engineer the A/D - D/A and
driver to work out this relationship and write card specific functions
to set the volume levels for the soundcard output.

If you are making changes to the volume controls, the most valuable
feature to the kind of work I am doing would be the ability to set the
volume level in volts individually for each channel. Whatever you do
please don't hide the actual analog gain settings behind software gain
settings, it would make my work very difficult.

Cheers,
//Anders



> On Mon, Oct 09, 2006 at 09:55:23AM -0400, Lee Revell wrote:
> > On Mon, 2006-10-09 at 09:45 -0400, Daniel Jacobowitz wrote:
> > > > alsactl save & restore should work for softvol.
> > > > The softvol mixer element doesn't exist only at the very fresh state.
> > > 
> > > I see, I just tried it and alsactl will create a user control just
> > > fine.  That's great.
> > > 
> > > I will give the softvol solution a try, and post the patches if it
> > > works.
> > > 
> > 
> > Isn't it inefficient to force all audio to go through a software volume
> > control just to have a master volume?  Wouldn't it be better to make a
> > ganged control in the driver (until the alsa-lib abstract mixer layer is
> > ready)?
> 
> I did consider doing it this way over the weekend.  However, being able
> to adjust the relative volumes of the different channels still seems
> useful to me.  If you have a single unified control, then you mess up
> trying to move the others.
> 
> What do you think?  Should I just slave them all together and remove
> the other volume controls?  It's probably easier; it just seems somehow
> lacking.




-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV