[PATCH 0/2] Add eSpeak

[PATCH 0/2] Add eSpeak

[Leo Famulari]

These patches provide the eSpeak software speech synthesizer [0].

I need advice on what audio system to configure it to use.

This patch configures it to use PulseAudio if it is available, and to
use PortAudio otherwise. Of course, since I have included PulseAudio as
an input, PulseAudio is always available and it starts a PulseAudio
server if one is not running [1].

The other option is to use only PortAudio (tested and works for me).

I guess the factors are:
1) Does GuixSD have a default audio setup that we should target? If
GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
integrated into that sytem.
2) Does this package, which launches PulseAudio, work for anyone on a
foreign distro?

Can GuixSD users with audio please test it out? As well as users on
foreign distros? You can do so like this:
`espeak 'hello world'`

[0]
http://espeak.sourceforge.net/

[1] This is actually not the expected behaviour and I am going to file a
bug. The Makefile reads "'runtime' uses pulseaudio if it is running,
else uses portaudio". Instead, it starts PulseAudio on demand.

Leo Famulari (2):
  gnu: Add sonic.
  gnu: Add espeak.

 gnu/packages/audio.scm | 86 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 86 insertions(+)

-- 
2.6.4

[PATCH 1/2] gnu: Add sonic.

[Leo Famulari]

* gnu/packages/audio.scm (sonic): New variable.
---
 gnu/packages/audio.scm | 34 ++++++++++++++++++++++++++++++++++
 1 file changed, 34 insertions(+)

diff --git a/gnu/packages/audio.scm b/gnu/packages/audio.scm
index ebae5d5..111a82d 100644
--- a/gnu/packages/audio.scm
+++ b/gnu/packages/audio.scm
@@ -2023,3 +2023,37 @@ that contains WAVE data, compressed or not---provided there exists a format
 module to handle that particular file type.")
     (home-page "http://etree.org/shnutils/shntool/")
     (license license:gpl3+)))
+
+(define-public sonic
+  (package
+    (name "sonic")
+    (version "0.2.0")
+    (source (origin
+             (method url-fetch)
+             (uri (string-append "https://github.com/waywardgeek/sonic/archive/"
+                                 "release-"
+                                 version ".tar.gz"))
+             (file-name (string-append name "-" version ".tar.gz"))
+             (sha256
+              (base32
+               "11a0q9wkgbb9ymf52v7dvybfhj8hprgr67zs1xcng143fvjpr0n7"))))
+    (build-system gnu-build-system)
+    (arguments
+      `(#:tests? #f ; no test suite
+        #:make-flags
+         (list (string-append "DESTDIR=" (assoc-ref %outputs "out")))
+        #:phases (modify-phases %standard-phases
+         (delete 'configure))))
+    (synopsis "Speed up or slow down speech")
+    (description "Sonic is a simple algorithm for speeding up or slowing down
+speech.  However, it's optimized for speed ups of over 2X, unlike previous
+algorithms for changing speech rate.  The Sonic library is a very simple ANSI C
+library that is designed to easily be integrated into streaming voice
+applications, like TTS back ends.
+
+The primary motivation behind Sonic is to enable the blind and visually impaired
+to improve their productivity with open source speech engines, like espeak.
+Sonic can also be used by the sighted.  For example, Sonic can improve the
+experience of listening to an audio book on an Android phone.")
+    (home-page "https://github.com/waywardgeek/sonic")
+    (license license:asl2.0)))
-- 
2.6.4

[PATCH 2/2] gnu: Add espeak.

[Leo Famulari]

* gnu/packages/audio.scm (espeak): New variable.
---
 gnu/packages/audio.scm | 52 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 52 insertions(+)

diff --git a/gnu/packages/audio.scm b/gnu/packages/audio.scm
index 111a82d..2f679b7 100644
--- a/gnu/packages/audio.scm
+++ b/gnu/packages/audio.scm
@@ -2057,3 +2057,55 @@ Sonic can also be used by the sighted.  For example, Sonic can improve the
 experience of listening to an audio book on an Android phone.")
     (home-page "https://github.com/waywardgeek/sonic")
     (license license:asl2.0)))
+
+(define-public espeak
+  (package
+    (name "espeak")
+    (version "1.48.04")
+    (source (origin
+             (method url-fetch)
+             (uri (string-append "mirror://sourceforge/project/espeak/espeak"
+                                 "/espeak-" (version-major+minor version)
+                                 "/espeak-" version "-source.zip"))
+             (sha256
+              (base32
+               "0n86gwh9pw0jqqpdz7mxggllfr8k0r7pc67ayy7w5z6z79kig6mz"))))
+    (build-system gnu-build-system)
+    (arguments
+      `(#:tests? #f ; no test suite
+        #:make-flags (let ((out (assoc-ref %outputs "out")))
+                       (list "--directory=src"
+                             ;; 'runtime' uses PulseAudio if it is running, else
+                             ;; it uses PortAudio.
+                             "AUDIO=runtime"
+                             ;; Add the 'lib' dir to the RUNPATH of the
+                             ;; binaries.
+                             (string-append "LDFLAGS=-Wl,-rpath=" out "/lib")
+                             (string-append "DATADIR=" out "/share/espeak-data")
+                             (string-append "PREFIX=" out)))
+        #:phases (modify-phases %standard-phases
+                   (add-after 'unpack 'patch-makefile
+                     (lambda _
+                       (substitute* "src/Makefile"
+                         (("LN_SF = /bin/ln -sf") "LN_SF = ln -sf"))))
+                    ;; The program supports portaudio versions 18 and 19. We
+                    ;; package 19, so we select it here.
+                   (add-after 'unpack 'pick-portaudio-version
+                     (lambda _
+                       (copy-file "src/portaudio19.h" "src/portaudio.h")))
+                   (delete 'configure))))
+    (native-inputs
+     `(("unzip" ,unzip)))
+    (inputs
+      `(("sonic" ,sonic)
+        ("pulseaudio" ,pulseaudio)
+        ("portaudio" ,portaudio)))
+    (synopsis "Software speech synthesizer")
+    (description "eSpeak is a compact software speech synthesizer for
+English and other languages, for Linux and Windows.  eSpeak uses a
+formant synthesis method.  This allows many languages to be provided in
+a small size.  The speech is clear, and can be used at high speeds, but
+is not as natural or smooth as larger synthesizers which are based on
+human speech recordings.")
+    (home-page "http://espeak.sourceforge.net/")
+    (license license:gpl3+)))
-- 
2.6.4

Re: [PATCH 0/2] Add eSpeak

[Leo Famulari]

On Tue, Jan 12, 2016 at 12:26:37AM -0500, Leo Famulari wrote:
> These patches provide the eSpeak software speech synthesizer [0].

I realized that the espeak upstream has gone inactive [0] and the users have
forked the project as espeak-ng: https://github.com/espeak-ng/espeak-ng/

The impression I get from the espeak ML is that the fork is merging a
lot of third-party patches that improve support for different languages,
as well as cleaning up the C codebase. So, in the future we should probably
package espeak-ng as well, for the sake of users that need speech
synthesis. It will conflict with espeak since the output binaries have
the same names.

There is also the espeakedit program that allows phoneme-editing. That
should be packaged, too.

[0] Read the last few months of their ML:
http://sourceforge.net/p/espeak/mailman/espeak-general/

> 
> I need advice on what audio system to configure it to use.
> 
> This patch configures it to use PulseAudio if it is available, and to
> use PortAudio otherwise. Of course, since I have included PulseAudio as
> an input, PulseAudio is always available and it starts a PulseAudio
> server if one is not running [1].
> 
> The other option is to use only PortAudio (tested and works for me).
> 
> I guess the factors are:
> 1) Does GuixSD have a default audio setup that we should target? If
> GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
> integrated into that sytem.
> 2) Does this package, which launches PulseAudio, work for anyone on a
> foreign distro?
> 
> Can GuixSD users with audio please test it out? As well as users on
> foreign distros? You can do so like this:
> `espeak 'hello world'`
> 
> [0]
> http://espeak.sourceforge.net/
> 
> [1] This is actually not the expected behaviour and I am going to file a
> bug. The Makefile reads "'runtime' uses pulseaudio if it is running,
> else uses portaudio". Instead, it starts PulseAudio on demand.
> 
> Leo Famulari (2):
>   gnu: Add sonic.
>   gnu: Add espeak.
> 
>  gnu/packages/audio.scm | 86 ++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 86 insertions(+)
> 
> -- 
> 2.6.4
> 
> 

Re: [PATCH 0/2] Add eSpeak

[Leo Famulari]

On Tue, Jan 12, 2016 at 12:58:12AM -0500, Leo Famulari wrote:
> On Tue, Jan 12, 2016 at 12:26:37AM -0500, Leo Famulari wrote:
> > These patches provide the eSpeak software speech synthesizer [0].
> 
> I realized that the espeak upstream has gone inactive [0] and the users have
> forked the project as espeak-ng: https://github.com/espeak-ng/espeak-ng/
> 
> The impression I get from the espeak ML is that the fork is merging a
> lot of third-party patches that improve support for different languages,
> as well as cleaning up the C codebase. So, in the future we should probably
> package espeak-ng as well, for the sake of users that need speech
> synthesis. It will conflict with espeak since the output binaries have
> the same names.

Or perhaps it will not conflict. I'm not sure. It's early days for
espeak-ng.

> 
> There is also the espeakedit program that allows phoneme-editing. That
> should be packaged, too.
> 
> [0] Read the last few months of their ML:
> http://sourceforge.net/p/espeak/mailman/espeak-general/
> 
> > 
> > I need advice on what audio system to configure it to use.
> > 
> > This patch configures it to use PulseAudio if it is available, and to
> > use PortAudio otherwise. Of course, since I have included PulseAudio as
> > an input, PulseAudio is always available and it starts a PulseAudio
> > server if one is not running [1].
> > 
> > The other option is to use only PortAudio (tested and works for me).
> > 
> > I guess the factors are:
> > 1) Does GuixSD have a default audio setup that we should target? If
> > GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
> > integrated into that sytem.
> > 2) Does this package, which launches PulseAudio, work for anyone on a
> > foreign distro?
> > 
> > Can GuixSD users with audio please test it out? As well as users on
> > foreign distros? You can do so like this:
> > `espeak 'hello world'`
> > 
> > [0]
> > http://espeak.sourceforge.net/
> > 
> > [1] This is actually not the expected behaviour and I am going to file a
> > bug. The Makefile reads "'runtime' uses pulseaudio if it is running,
> > else uses portaudio". Instead, it starts PulseAudio on demand.
> > 
> > Leo Famulari (2):
> >   gnu: Add sonic.
> >   gnu: Add espeak.
> > 
> >  gnu/packages/audio.scm | 86 ++++++++++++++++++++++++++++++++++++++++++++++++++
> >  1 file changed, 86 insertions(+)
> > 
> > -- 
> > 2.6.4
> > 
> > 
> 

Re: [PATCH 0/2] Add eSpeak

[Andreas Enge]

On Tue, Jan 12, 2016 at 12:58:12AM -0500, Leo Famulari wrote:
> So, in the future we should probably
> package espeak-ng as well, for the sake of users that need speech
> synthesis. It will conflict with espeak since the output binaries have
> the same names.

From what you write, would it not be better to directly package espeak-ng
and to drop espeak?

Andreas

Re: [PATCH 0/2] Add eSpeak

[Leo Famulari]

On Tue, Jan 12, 2016 at 04:48:18PM +0100, Andreas Enge wrote:
> On Tue, Jan 12, 2016 at 12:58:12AM -0500, Leo Famulari wrote:
> > So, in the future we should probably
> > package espeak-ng as well, for the sake of users that need speech
> > synthesis. It will conflict with espeak since the output binaries have
> > the same names.
> 
> From what you write, would it not be better to directly package espeak-ng
> and to drop espeak?

Eventually we should do that, but for now I think we should provide the
espeak that users are familiar with. I am not sure how much the
interface has changed in espeak-ng, if at all. I'm going to try it out
soon.

> 
> Andreas
> 

Re: [PATCH 0/2] Add eSpeak

[Leo Famulari]

On Tue, Jan 12, 2016 at 02:30:06PM -0500, Leo Famulari wrote:
> On Tue, Jan 12, 2016 at 04:48:18PM +0100, Andreas Enge wrote:
> > On Tue, Jan 12, 2016 at 12:58:12AM -0500, Leo Famulari wrote:
> > > So, in the future we should probably
> > > package espeak-ng as well, for the sake of users that need speech
> > > synthesis. It will conflict with espeak since the output binaries have
> > > the same names.
> > 
> > From what you write, would it not be better to directly package espeak-ng
> > and to drop espeak?
> 
> Eventually we should do that, but for now I think we should provide the
> espeak that users are familiar with. I am not sure how much the
> interface has changed in espeak-ng, if at all. I'm going to try it out
> soon.

There's a WIP patch for espeak-ng available here if anyone wants to take
a crack at making it work:
https://github.com/lfam/guix/commits/contrib-espeak

I think the error it fails on is corrected on the git master branch.
I'll try it soon.

> 
> > 
> > Andreas
> > 
> 

PulseAudio

[Ludovic Courtès]

Leo Famulari <leo@famulari.name> skribis:

> I guess the factors are:
> 1) Does GuixSD have a default audio setup that we should target? If
> GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
> integrated into that sytem.
> 2) Does this package, which launches PulseAudio, work for anyone on a
> foreign distro?

It’s not written anywhere, but I think most of our audio packages target
PulseAudio (that’s what I use on GuixSD.)  I’m in favor of consistently
using it, and it would probably be best to write it down in the manual.

Thoughts?

Ludo’.

Re: [PATCH 1/2] gnu: Add sonic.

[Ludovic Courtès]

Leo Famulari <leo@famulari.name> skribis:

> * gnu/packages/audio.scm (sonic): New variable.

[...]

> +        #:phases (modify-phases %standard-phases
> +         (delete 'configure))))

There’s a problem here.  :-)

> +    (synopsis "Speed up or slow down speech")
> +    (description "Sonic is a simple algorithm for speeding up or slowing down

s/is a/implements a/

> +speech.  However, it's optimized for speed ups of over 2X, unlike previous
> +algorithms for changing speech rate.  The Sonic library is a very simple ANSI C
> +library that is designed to easily be integrated into streaming voice
> +applications, like TTS back ends.

I would change the last sentence to:

  Sonic is a C library designed to be easily integrated into streaming
  voice applications such as text-to-speech (TTS) back ends.

> +The primary motivation behind Sonic is to enable the blind and visually impaired
> +to improve their productivity with open source speech engines, like espeak.
                                ^‑------------------------------------------^
Replace with “with speech engines such as eSpeak.”

> +Sonic can also be used by the sighted.  For example, Sonic can improve the
> +experience of listening to an audio book on an Android phone.")

I’d remove these two sentences.

OK with these changes.

Thanks,
Ludo’.

Re: [PATCH 2/2] gnu: Add espeak.

[Ludovic Courtès]

Leo Famulari <leo@famulari.name> skribis:

> * gnu/packages/audio.scm (espeak): New variable.

If you think eSpeak-ng is not ready yet, or is sufficiently different,
it’s OK to package eSpeak.

> +    (native-inputs
> +     `(("unzip" ,unzip)))
> +    (inputs
> +      `(("sonic" ,sonic)

The “unzip” line is misaligned.

> +        ("pulseaudio" ,pulseaudio)
> +        ("portaudio" ,portaudio)))

Dunno if it’s useful to have both, maybe someone more knowledgeable like
Ricardo can answer?

> +    (synopsis "Software speech synthesizer")
> +    (description "eSpeak is a compact software speech synthesizer for
> +English and other languages, for Linux and Windows.  eSpeak uses a
                              ^---------------------^
Remove.

In general, we don’t mention proprietary software, and we don’t mention
other OSes, free or not, portability etc., because that’s not what
matters from the viewpoint of an OS provider.

The rest LGTM!

Thanks,
Ludo’.

Re: [PATCH 2/2] gnu: Add espeak.

[Leo Famulari]

On Wed, Jan 13, 2016 at 03:17:28PM +0100, Ludovic Courtès wrote:
> Leo Famulari <leo@famulari.name> skribis:
> 
> > * gnu/packages/audio.scm (espeak): New variable.
> 
> If you think eSpeak-ng is not ready yet, or is sufficiently different,
> it’s OK to package eSpeak.

I'm looking into it.

> 
> > +    (native-inputs
> > +     `(("unzip" ,unzip)))
> > +    (inputs
> > +      `(("sonic" ,sonic)
> 
> The “unzip” line is misaligned.

The "unzip" line or the "sonic" line? Most of the patches I see use the
former style.

> 
> > +        ("pulseaudio" ,pulseaudio)
> > +        ("portaudio" ,portaudio)))
> 
> Dunno if it’s useful to have both, maybe someone more knowledgeable like
> Ricardo can answer?

I misinterpreted the configure options. It's not useful to have both. If
PulseAudio is an input, then eSpeak will always use PulseAudio unless
configured not to, in which case PulseAudio should not be an input.

PulseAudio stopped working completely on my Debian system a while ago.
I decided to keep using ALSA and never looked back (or is it forward, in
this case? ;)

So, if we are going to target PulseAudio, can a PulseAudio user test
that this package works? This should make the computer speak "hello
world":
$ espeak "hello world"

> 
> > +    (synopsis "Software speech synthesizer")
> > +    (description "eSpeak is a compact software speech synthesizer for
> > +English and other languages, for Linux and Windows.  eSpeak uses a
>                               ^---------------------^
> Remove.

Done.

> 
> In general, we don’t mention proprietary software, and we don’t mention
> other OSes, free or not, portability etc., because that’s not what
> matters from the viewpoint of an OS provider.
> 
> The rest LGTM!

Again, I'll be grateful if a PulseAudio user will test!

> 
> Thanks,
> Ludo’.

Re: [PATCH 1/2] gnu: Add sonic.

[Leo Famulari]

On Wed, Jan 13, 2016 at 03:10:16PM +0100, Ludovic Courtès wrote:
> Leo Famulari <leo@famulari.name> skribis:
> 
> > * gnu/packages/audio.scm (sonic): New variable.
> 
> [...]
> 
> > +        #:phases (modify-phases %standard-phases
> > +         (delete 'configure))))
              *-------->|
> 
> There’s a problem here.  :-)

The indentation of the second line, correct?

> 
> > +    (synopsis "Speed up or slow down speech")
> > +    (description "Sonic is a simple algorithm for speeding up or slowing down
> 
> s/is a/implements a/

> 
> > +speech.  However, it's optimized for speed ups of over 2X, unlike previous
> > +algorithms for changing speech rate.  The Sonic library is a very simple ANSI C
> > +library that is designed to easily be integrated into streaming voice
> > +applications, like TTS back ends.
> 
> I would change the last sentence to:
> 
>   Sonic is a C library designed to be easily integrated into streaming
>   voice applications such as text-to-speech (TTS) back ends.
> 
> > +The primary motivation behind Sonic is to enable the blind and visually impaired
> > +to improve their productivity with open source speech engines, like espeak.
>                                 ^‑------------------------------------------^
> Replace with “with speech engines such as eSpeak.”
> 
> > +Sonic can also be used by the sighted.  For example, Sonic can improve the
> > +experience of listening to an audio book on an Android phone.")
> 
> I’d remove these two sentences.
> 
> OK with these changes.

Done.

> 
> Thanks,
> Ludo’.

Re: [PATCH 2/2] gnu: Add espeak.

[Ricardo Wurmus]

Ludovic Courtès <ludo@gnu.org> writes:

> Leo Famulari <leo@famulari.name> skribis:
>
>> * gnu/packages/audio.scm (espeak): New variable.
>
> If you think eSpeak-ng is not ready yet, or is sufficiently different,
> it’s OK to package eSpeak.
>
>> +    (native-inputs
>> +     `(("unzip" ,unzip)))
>> +    (inputs
>> +      `(("sonic" ,sonic)
>
> The “unzip” line is misaligned.
>
>> +        ("pulseaudio" ,pulseaudio)
>> +        ("portaudio" ,portaudio)))
>
> Dunno if it’s useful to have both, maybe someone more knowledgeable like
> Ricardo can answer?

It *might* be useful for some people to have portaudio support, but we
don’t usually add all supported audio backends.  We don’t even always
add pulseaudio to the inputs, because pulseaudio can readily deal with
programmes that only use the ALSA backend.

I’m not using pulseaudio; for me it’s just ALSA for regular software +
JACK for “professional” audio where sync and timing matters.

I’m not very knowledgeable about pulseaudio, unfortunately.  As
pulseaudio can manage audio streams using various different backends
from pulseaudio-agnostic software, what really is to be gained by adding
pulseaudio to the inputs?  AFAIK using pulseaudio directly is not much
different from using ALSA and configuring pulseaudio to manage this
stream.

~~ Ricardo

Re: [PATCH 2/2] gnu: Add espeak.

[Ludovic Courtès]

Ricardo Wurmus <ricardo.wurmus@mdc-berlin.de> skribis:

> I’m not using pulseaudio; for me it’s just ALSA for regular software +
> JACK for “professional” audio where sync and timing matters.

That’s what I expected.  ;-)

> I’m not very knowledgeable about pulseaudio, unfortunately.  As
> pulseaudio can manage audio streams using various different backends
> from pulseaudio-agnostic software, what really is to be gained by adding
> pulseaudio to the inputs?  AFAIK using pulseaudio directly is not much
> different from using ALSA and configuring pulseaudio to manage this
> stream.

As you note, the main difference is configuration.  For ALSA-lib to use
PulseAudio, one has to drop the relevant ~/.asoundrc, but it seems that
it does not always work properly.

For instance, my ~/.asoundrc reads this:

--8<---------------cut here---------------start------------->8---
# FIXME: alsamixer & co. from alsa-utils don't support it.

pcm.!default {
    type pulse
}

# ctl.!default {
#     type pulse
# }
--8<---------------cut here---------------end--------------->8---

I don’t recall the details of the FIXME and the commented-out part,
though.

For GuixSD I think it would be best if everything would automatically go
through PulseAudio, without the user having to configure obscure things.

Ludo’.

Re: [PATCH 2/2] gnu: Add espeak.

[Leo Famulari]

On Wed, Jan 13, 2016 at 10:27:13PM -0500, Leo Famulari wrote:
> On Wed, Jan 13, 2016 at 03:17:28PM +0100, Ludovic Courtès wrote:
> > Leo Famulari <leo@famulari.name> skribis:
> > 
> > > * gnu/packages/audio.scm (espeak): New variable.
> > 
> > If you think eSpeak-ng is not ready yet, or is sufficiently different,
> > it’s OK to package eSpeak.
> 
> I'm looking into it.

I think it will be a great replacement for eSpeak, but it's not ready
yet, in my opinion. The new autotools-based build system has a lot of
kinks to work out and recent git HEADs don't build at all for me. When
it is ready we should package it.

> 
> > 
> > > +    (native-inputs
> > > +     `(("unzip" ,unzip)))
> > > +    (inputs
> > > +      `(("sonic" ,sonic)
> > 
> > The “unzip” line is misaligned.
> 
> The "unzip" line or the "sonic" line? Most of the patches I see use the
> former style.
> 
> > 
> > > +        ("pulseaudio" ,pulseaudio)
> > > +        ("portaudio" ,portaudio)))
> > 
> > Dunno if it’s useful to have both, maybe someone more knowledgeable like
> > Ricardo can answer?
> 
> I misinterpreted the configure options. It's not useful to have both. If
> PulseAudio is an input, then eSpeak will always use PulseAudio unless
> configured not to, in which case PulseAudio should not be an input.
> 
> PulseAudio stopped working completely on my Debian system a while ago.
> I decided to keep using ALSA and never looked back (or is it forward, in
> this case? ;)
> 
> So, if we are going to target PulseAudio, can a PulseAudio user test
> that this package works? This should make the computer speak "hello
> world":
> $ espeak "hello world"
> 
> > 
> > > +    (synopsis "Software speech synthesizer")
> > > +    (description "eSpeak is a compact software speech synthesizer for
> > > +English and other languages, for Linux and Windows.  eSpeak uses a
> >                               ^---------------------^
> > Remove.
> 
> Done.
> 
> > 
> > In general, we don’t mention proprietary software, and we don’t mention
> > other OSes, free or not, portability etc., because that’s not what
> > matters from the viewpoint of an OS provider.
> > 
> > The rest LGTM!
> 
> Again, I'll be grateful if a PulseAudio user will test!
> 
> > 
> > Thanks,
> > Ludo’.
> 

Re: [PATCH 2/2] gnu: Add espeak.

[Ludovic Courtès]

Leo Famulari <leo@famulari.name> skribis:

> On Wed, Jan 13, 2016 at 10:27:13PM -0500, Leo Famulari wrote:
>> On Wed, Jan 13, 2016 at 03:17:28PM +0100, Ludovic Courtès wrote:
>> > Leo Famulari <leo@famulari.name> skribis:
>> > 
>> > > * gnu/packages/audio.scm (espeak): New variable.
>> > 
>> > If you think eSpeak-ng is not ready yet, or is sufficiently different,
>> > it’s OK to package eSpeak.
>> 
>> I'm looking into it.
>
> I think it will be a great replacement for eSpeak, but it's not ready
> yet, in my opinion.

OK, so that’s a good reason to package eSpeak!

Thanks,
Ludo’.

Re: [PATCH 2/2] gnu: Add espeak.

[Ricardo Wurmus]

Ludovic Courtès <ludo@gnu.org> writes:

> Ricardo Wurmus <ricardo.wurmus@mdc-berlin.de> skribis:
>
>> I’m not using pulseaudio; for me it’s just ALSA for regular software +
>> JACK for “professional” audio where sync and timing matters.
>
> That’s what I expected.  ;-)
>
>> I’m not very knowledgeable about pulseaudio, unfortunately.  As
>> pulseaudio can manage audio streams using various different backends
>> from pulseaudio-agnostic software, what really is to be gained by adding
>> pulseaudio to the inputs?  AFAIK using pulseaudio directly is not much
>> different from using ALSA and configuring pulseaudio to manage this
>> stream.
>
> As you note, the main difference is configuration.  For ALSA-lib to use
> PulseAudio, one has to drop the relevant ~/.asoundrc, but it seems that
> it does not always work properly.
>
> For instance, my ~/.asoundrc reads this:
>
> --8<---------------cut here---------------start------------->8---
> # FIXME: alsamixer & co. from alsa-utils don't support it.
>
> pcm.!default {
>     type pulse
> }
>
> # ctl.!default {
> #     type pulse
> # }
> --8<---------------cut here---------------end--------------->8---
>
> I don’t recall the details of the FIXME and the commented-out part,
> though.
>
> For GuixSD I think it would be best if everything would automatically go
> through PulseAudio, without the user having to configure obscure things.

Couldn’t we then just install the appropriate /etc/asound.conf (maybe
with a “pulseaudio-alsa” service)?  Users could still overwrite it by
disabling the service or with “~/.asoundrc”.

A similar configuration is provided on Arch by the “pulseaudio-alsa”
package, which should be sufficient to route all ALSA audio streams
through Pulseaudio.

Wouldn’t that be a better solution than to add the pulseaudio package to
the inputs wherever possible?

~~ Ricardo

PulseAudio & ALSA

[Ludovic Courtès]

Ricardo Wurmus <rekado@elephly.net> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> Ricardo Wurmus <ricardo.wurmus@mdc-berlin.de> skribis:
>>
>>> I’m not using pulseaudio; for me it’s just ALSA for regular software +
>>> JACK for “professional” audio where sync and timing matters.
>>
>> That’s what I expected.  ;-)
>>
>>> I’m not very knowledgeable about pulseaudio, unfortunately.  As
>>> pulseaudio can manage audio streams using various different backends
>>> from pulseaudio-agnostic software, what really is to be gained by adding
>>> pulseaudio to the inputs?  AFAIK using pulseaudio directly is not much
>>> different from using ALSA and configuring pulseaudio to manage this
>>> stream.
>>
>> As you note, the main difference is configuration.  For ALSA-lib to use
>> PulseAudio, one has to drop the relevant ~/.asoundrc, but it seems that
>> it does not always work properly.
>>
>> For instance, my ~/.asoundrc reads this:
>>
>> --8<---------------cut here---------------start------------->8---
>> # FIXME: alsamixer & co. from alsa-utils don't support it.
>>
>> pcm.!default {
>>     type pulse
>> }
>>
>> # ctl.!default {
>> #     type pulse
>> # }
>> --8<---------------cut here---------------end--------------->8---
>>
>> I don’t recall the details of the FIXME and the commented-out part,
>> though.
>>
>> For GuixSD I think it would be best if everything would automatically go
>> through PulseAudio, without the user having to configure obscure things.
>
> Couldn’t we then just install the appropriate /etc/asound.conf (maybe
> with a “pulseaudio-alsa” service)?  Users could still overwrite it by
> disabling the service or with “~/.asoundrc”.
>
> A similar configuration is provided on Arch by the “pulseaudio-alsa”
> package, which should be sufficient to route all ALSA audio streams
> through Pulseaudio.

Sounds like a good idea.

> Wouldn’t that be a better solution than to add the pulseaudio package to
> the inputs wherever possible?

Dunno.  It seems to me that applications that are PA-aware can have
better integration, like have their name shown in pavucontrol and
similar interfaces, which is quite useful.

There are also technical complications: alsa-lib must be able to find
its PulseAudio plug-in (currently we don’t build it since alsa-lib would
depend on PulseAudio), and it must be one that can actually be dlopened
by every process out there (linked against the same libc, etc.)

Thanks,
Ludo’.

Re: PulseAudio

[Christopher Allan Webber]

Ludovic Courtès writes:

> Leo Famulari <leo@famulari.name> skribis:
>
>> I guess the factors are:
>> 1) Does GuixSD have a default audio setup that we should target? If
>> GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
>> integrated into that sytem.
>> 2) Does this package, which launches PulseAudio, work for anyone on a
>> foreign distro?
>
> It’s not written anywhere, but I think most of our audio packages target
> PulseAudio (that’s what I use on GuixSD.)  I’m in favor of consistently
> using it, and it would probably be best to write it down in the manual.
>
> Thoughts?
>
> Ludo’.

I'd really like it if we just agreed that in general, yeah, we want
Pulseaudio support.  I used to spend all sorts of time fighting my audio
setup, being careful on what application opened first so I could be sure
which one grabbed control of Alsa.  I hadn't had this problem in a few
years, and it was so nice not to worry about it, but on Guix I've had
some times where I have again... eg playing audio/video in Icecat
seems not to use it for whatever reason, and it's a bummer, because then
I have to be careful about when I start rhythmbox or whatever.

I'd love to mostly not worry/think about audio issues again!

Re: PulseAudio

[Leo Famulari]

On Tue, Feb 09, 2016 at 01:26:32PM -0800, Christopher Allan Webber wrote:
> Ludovic Courtès writes:
> 
> > Leo Famulari <leo@famulari.name> skribis:
> >
> >> I guess the factors are:
> >> 1) Does GuixSD have a default audio setup that we should target? If
> >> GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
> >> integrated into that sytem.
> >> 2) Does this package, which launches PulseAudio, work for anyone on a
> >> foreign distro?
> >
> > It’s not written anywhere, but I think most of our audio packages target
> > PulseAudio (that’s what I use on GuixSD.)  I’m in favor of consistently
> > using it, and it would probably be best to write it down in the manual.
> >
> > Thoughts?
> >
> > Ludo’.
> 
> I'd really like it if we just agreed that in general, yeah, we want
> Pulseaudio support.  I used to spend all sorts of time fighting my audio
> setup, being careful on what application opened first so I could be sure
> which one grabbed control of Alsa.  I hadn't had this problem in a few
> years, and it was so nice not to worry about it, but on Guix I've had
> some times where I have again... eg playing audio/video in Icecat
> seems not to use it for whatever reason, and it's a bummer, because then
> I have to be careful about when I start rhythmbox or whatever.
> 
> I'd love to mostly not worry/think about audio issues again!

Heh, that's why I uninstall PulseAudio every time a Debian upgrade
brings it in ;)

All kidding aside, I'm happy to push this with PulseAudio support, *if*
a GuixSD user can confirm that it works. I don't have sound on my GuixSD
installation.

Re: PulseAudio

[Ricardo Wurmus]

Christopher Allan Webber <cwebber@dustycloud.org> writes:

> Ludovic Courtès writes:
>
>> Leo Famulari <leo@famulari.name> skribis:
>>
>>> I guess the factors are:
>>> 1) Does GuixSD have a default audio setup that we should target? If
>>> GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
>>> integrated into that sytem.
>>> 2) Does this package, which launches PulseAudio, work for anyone on a
>>> foreign distro?
>>
>> It’s not written anywhere, but I think most of our audio packages target
>> PulseAudio (that’s what I use on GuixSD.)  I’m in favor of consistently
>> using it, and it would probably be best to write it down in the manual.
>>
>> Thoughts?
>>
>> Ludo’.
>
> I'd really like it if we just agreed that in general, yeah, we want
> Pulseaudio support.

I agree with Pulseaudio *support*, but this does not mean that we’d have
to link all applications that have an optional Pulseaudio backend with
the Pulseaudio libraries, does it?

Since I’m not using Pulseaudio I cannot vouch for this to work, but ALSA
can be configured to use redirect audio streams to Pulseaudio (if it’s
running).  I know of no application that offers a Pulseaudio backend but
cannot use ALSA directly.  This redirection from ALSA to Pulseaudio
could be implemented as an etc-service, providing the required
configuration file.

My experience with Pulseaudio was very different (and still is if I
count support requests from friends), but maybe that’s just because I
don’t *always* want Pulseaudio.  I found it difficult to reliably
disable/suspend Pulseaudio whenever I wanted to do audio work (=
involving JACK).  It often just got in the way.

On other machines where I didn’t use JACK regularly Pulseaudio was
acceptable, I guess.  It makes me shed a salty tear that something so
complicated and layered as Pulseaudio was necessary to tame the mess in
the Linux [sic] audio world :-/

~~ Ricardo

Re: PulseAudio & ALSA

[Ricardo Wurmus]

Ludovic Courtès <ludo@gnu.org> writes:

>> Wouldn’t that be a better solution than to add the pulseaudio package to
>> the inputs wherever possible?
>
> Dunno.  It seems to me that applications that are PA-aware can have
> better integration, like have their name shown in pavucontrol and
> similar interfaces, which is quite useful.

I take your word for it.  I haven’t noticed.

> There are also technical complications: alsa-lib must be able to find
> its PulseAudio plug-in (currently we don’t build it since alsa-lib would
> depend on PulseAudio), and it must be one that can actually be dlopened
> by every process out there (linked against the same libc, etc.)

Hmm, that does sound hairy.

~~ Ricardo

Re: PulseAudio

[Jookia]

On Tue, Feb 09, 2016 at 01:26:32PM -0800, Christopher Allan Webber wrote:
> Ludovic Courtès writes:
> 
> > Leo Famulari <leo@famulari.name> skribis:
> >
> >> I guess the factors are:
> >> 1) Does GuixSD have a default audio setup that we should target? If
> >> GuixSD uses PulseAudio, then I think it would be good for eSpeak to be
> >> integrated into that sytem.
> >> 2) Does this package, which launches PulseAudio, work for anyone on a
> >> foreign distro?
> >
> > It’s not written anywhere, but I think most of our audio packages target
> > PulseAudio (that’s what I use on GuixSD.)  I’m in favor of consistently
> > using it, and it would probably be best to write it down in the manual.
> >
> > Thoughts?
> >
> > Ludo’.
> 
> I'd really like it if we just agreed that in general, yeah, we want
> Pulseaudio support.  I used to spend all sorts of time fighting my audio
> setup, being careful on what application opened first so I could be sure
> which one grabbed control of Alsa.  I hadn't had this problem in a few
> years, and it was so nice not to worry about it, but on Guix I've had
> some times where I have again... eg playing audio/video in Icecat
> seems not to use it for whatever reason, and it's a bummer, because then
> I have to be careful about when I start rhythmbox or whatever.

> I'd love to mostly not worry/think about audio issues again!

Going to chime in here: While I don't like PulseAudio, it's a requirement for
working audio on the Novena which I plan to run GuixSD on. Something something
necessary evil. That said, if there's a big rebuild planned there is a
PulseAudio patch I needed to make for NixOS to get things working right. I'm
still tinkering with early boot issues in GuixSD so I'm not that far in the
stack to test it, but if anyone's curious I'll attach the NixOS patch here.

Jookia.

Re: PulseAudio

[Ludovic Courtès]

Leo Famulari <leo@famulari.name> skribis:

> All kidding aside, I'm happy to push this with PulseAudio support, *if*
> a GuixSD user can confirm that it works. I don't have sound on my GuixSD
> installation.

I use it all the time and it works, except for IceCat & co. as Chris
notes, which try to use ALSA directly.  The person who fixes will make
at least two people happy, and that’s really worth it!  :-)

Ludo’.

Re: PulseAudio

[Ricardo Wurmus]

Ludovic Courtès <ludo@gnu.org> writes:

> Leo Famulari <leo@famulari.name> skribis:
>
>> All kidding aside, I'm happy to push this with PulseAudio support, *if*
>> a GuixSD user can confirm that it works. I don't have sound on my GuixSD
>> installation.
>
> I use it all the time and it works, except for IceCat & co. as Chris
> notes, which try to use ALSA directly.

Does this mean you do not hear sound coming from IceCat and its plugins
with the example desktop system configuration?

~~ Ricardo

Re: PulseAudio

[Alex Sassmannshausen]

Hello,

Ricardo Wurmus writes:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> Leo Famulari <leo@famulari.name> skribis:
>>
>>> All kidding aside, I'm happy to push this with PulseAudio support, *if*
>>> a GuixSD user can confirm that it works. I don't have sound on my GuixSD
>>> installation.
>>
>> I use it all the time and it works, except for IceCat & co. as Chris
>> notes, which try to use ALSA directly.
>
> Does this mean you do not hear sound coming from IceCat and its plugins
> with the example desktop system configuration?

This is my experience too: works lovely for things like VLC, but not so
much for IceCat.

Alex

Re: PulseAudio

[Ricardo Wurmus]

Alex Sassmannshausen <alex.sassmannshausen@gmail.com> writes:

> Hello,
>
> Ricardo Wurmus writes:
>
>> Ludovic Courtès <ludo@gnu.org> writes:
>>
>>> Leo Famulari <leo@famulari.name> skribis:
>>>
>>>> All kidding aside, I'm happy to push this with PulseAudio support, *if*
>>>> a GuixSD user can confirm that it works. I don't have sound on my GuixSD
>>>> installation.
>>>
>>> I use it all the time and it works, except for IceCat & co. as Chris
>>> notes, which try to use ALSA directly.
>>
>> Does this mean you do not hear sound coming from IceCat and its plugins
>> with the example desktop system configuration?
>
> This is my experience too: works lovely for things like VLC, but not so
> much for IceCat.

How strange!  Do you have a configuration file “~/.asoundrc” or
“/etc/asound.conf” that reroutes ALSA to PulseAudio?

I’m not using PulseAudio on GuixSD and I have sound in media players and
browser alike.

~~ Ricardo

PS: don’t think I’m blocking the use of PulseAudio.  If it works with PA
when built with the PA libs (and doesn’t work without it), then it’s
fine to make a dependency on the PA libs, of course.  I’m just curious
to learn why rerouting ALSA to PA doesn’t work — because rerouting is
the recommended way to get plain ALSA programmes to work nice with PA.

Re: PulseAudio

[Alex Sassmannshausen]

Hiya,

Ricardo Wurmus writes:

> Alex Sassmannshausen <alex.sassmannshausen@gmail.com> writes:
>
>> Hello,
>>
>> Ricardo Wurmus writes:
>>
>>> Ludovic Courtès <ludo@gnu.org> writes:
>>>
>>>> Leo Famulari <leo@famulari.name> skribis:
>>>>
>>>>> All kidding aside, I'm happy to push this with PulseAudio support, *if*
>>>>> a GuixSD user can confirm that it works. I don't have sound on my GuixSD
>>>>> installation.
>>>>
>>>> I use it all the time and it works, except for IceCat & co. as Chris
>>>> notes, which try to use ALSA directly.
>>>
>>> Does this mean you do not hear sound coming from IceCat and its plugins
>>> with the example desktop system configuration?
>>
>> This is my experience too: works lovely for things like VLC, but not so
>> much for IceCat.
>
> How strange!  Do you have a configuration file “~/.asoundrc” or
> “/etc/asound.conf” that reroutes ALSA to PulseAudio?
>
> I’m not using PulseAudio on GuixSD and I have sound in media players and
> browser alike.

Sorry — I think I misunderstood what is going on: seems I don't have
PulseAudio enabled on my system at the moment — please ignore the noise.

Still curious why I don't have sound in my browser… more investigation
on my side needed it seems.

Cheers,

Alex