Is there a project that is jovie successor? (fwd)
Jeremy Whiting
jpwhiting at kde.org
Wed Oct 27 20:48:28 BST 2021
Hi Gustav,
That looks a lot more interesting. So just to confirm the mary had a lamb
lines sound fine on the speaker/headphones but the larger text from pdf via
okular still sounds strange or as if there are multiple things being said
at the same time?
I don't see much difference between the 2, I tried sending longer text
through okular here and got some weird different accent somehow, but see
similar things in espeak.log about it selecting en+0 like you have there. I
think it's possible espeak is switching languages for some reason, maybe
when it gets non utf8 text like the character between the copyright symbol
"Authors ... Contributions" but I tried having spd-say read just that and
it read it fine.
I then put a whole block of text from your log like this into a
testfile.txt and use this to read:
testfile.txt contents:
<speak>Statement of Ethics
Written informed consent was obtained from the patient for
publication of this case report and any accompanying images.<mark
name="__spd_0"/>
Authors□~@~Y Contributions
Elena Rudakova was the primary physician who treated the
patient and who obtained and analyzed the data for the study.<mark
name="__spd_1"/>
Seema Mahesh analyzed the data, wrote the manuscript, and
obtained the references.<mark name="__spd_2"/> George Vithoulkas was the
guide,
final approver, and guarantor of the work.<mark name="__spd_3"/> All
authors have
read and approved the final version of the manuscript.<mark name="__spd_4"/>
Acknowledgements
The authors are thankful to Aishwarya Madhusudhan for the
technical support.<mark name="__spd_5"/>
</speak>
cat ./testfile.txt | spd-say -x -e
-x makes it parse the <speak> and <mark> tags, etc. and -e makes it read
text given from standard input that cat is giving it.
When I tried that here it read about a line, then got an assertion.
>From my espeak.log here:
Wed Oct 27 13:28:19 2021 [333272]: Espeak: module_speak().
Wed Oct 27 13:28:19 2021 [333297]: Espeak: Requested data: |publication of
this case report and any accompanying images.<mark name="__spd_0"/><mark
name="__spd_0"/>
| 0 105
Wed Oct 27 13:28:19 2021 [384129]: Espeak: Leaving module_speak() normally.
Wed Oct 27 13:28:19 2021 [385487]: Espeak: Playback semaphore on.
Wed Oct 27 13:28:19 2021 [385499]: Printing reply: 701 BEGIN
Wed Oct 27 13:28:19 2021 [385509]: Printed
Wed Oct 27 13:28:19 2021 [385518]: Espeak: Sending 51283 samples to audio.
Wed Oct 27 13:28:20 2021 [878168]: CMD: <STOP
>
Wed Oct 27 13:28:20 2021 [878194]: Espeak: module_stop().
Wed Oct 27 13:28:20 2021 [878217]: Espeak: stopping...
Wed Oct 27 13:28:20 2021 [878237]: CMD: <QUIT
>
Wed Oct 27 13:28:20 2021 [878278]: Espeak: Stop or pause semaphore on.
Wed Oct 27 13:28:20 2021 [878299]: Espeak: Stopping audio.
Assertion 'pthread_mutex_destroy(&m->mutex) == 0' failed at
pulsecore/mutex-posix.c:83, function pa_mutex_free(). Aborting.
I think espeak is having issues with the longer text given from okular.
I took a quick look at using espeak-ng on opensuse 15.2, but that doesn't
seem to be possible, nothing is providing sd_espeak-ng speech-dispatcher
module from the available packages which would be needed to use espeak-ng
instead of espeak.
If you create a text file like above and cat it into spd-say -x -e like
above does it duplicate then also?
BR,
Jeremy
On Wed, Oct 27, 2021 at 10:41 AM Gustav Degreef <gustav97 at gmail.com> wrote:
> Sorry Jeremy. I rushed the previous response and did not see the last
> part regarding the Debug 1. Yes, I had to change the Debug from 0 to 1.
> I rebooted to make sure both espeak and speech-dispatcher used the
> updated config files. I had spd-say speak 3 sentences and I had Okular
> read 1/2 page of a pdf three times, just in case. Here are the zipped
> logs. I deleted the old logs before rebooting. Thanks for your patient
> efforts, Gustav.
>
> On 10/26/21 8:00 PM, Jeremy Whiting wrote:
> > Gustav,
> >
> > Yeah, I think the missing bits were the Debug 1 in espeak.conf file. I
> > had that here but forgot to mention it originally. Also, the extra
> > details are in the espeak.log not speech-dispatcher.log it seems. The
> > one you sent was empty, but that's probably because of the missing
> > Debug 1 in there. Once that's in place it should give a bunch of this
> > kind of details.
> >
> > thanks,
> > Jeremy
> >
> > On Tue, Oct 26, 2021 at 11:41 AM Gustav Degreef <gustav97 at gmail.com
> > <mailto:gustav97 at gmail.com>> wrote:
> >
> > Hi Jeremy,
> >
> > I did EXACTLY as you said when I sent you the zip file. When I
> > first
> > edited the .conf file, there was no output to the logs, so I rebooted
> > the system (I did not know how to re-start the deamon). No clue
> > why the
> > output you expected is not there. Â So now I did the same thing
> > again,
> > but I had Okular read two separate long instnaces and then had
> > spd-say a
> > whole sentence twice. Here is the single log file (I did not
> > delete the
> > old log file, so I figure the new info has been appended. Gustav
> >
> > On 10/26/21 7:03 PM, Jeremy Whiting wrote:
> > > Gustav,
> > >
> > > I see speech dispatcher itself started and ended a couple of
> > times in
> > > there, but I don't see what text it was sent to synthesize. Did you
> > > try spd-say "Hello" or trying with okular to speak some document
> > or a
> > > page? I was hoping to see something like this:
> > >
> > > Tue Oct 26 09:31:25 2021 [617497]: Espeak: Stop or pause thread
> > > starting.......
> > >  Tue Oct 26 09:31:25 2021 [636516]: CMD: <SET
> > > >
> > >  Tue Oct 26 09:31:25 2021 [636587]: CMD: <SPEAK
> > > >
> > >  Tue Oct 26 09:31:25 2021 [636617]: Espeak: module_speak().
> > >  Tue Oct 26 09:31:25 2021 [636627]: Espeak: Requested data:
> > > |<speak>hello</speak>| 0 20
> > >  Tue Oct 26 09:31:25 2021 [636637]: Espeak:
> > set_language_and_voice en 1
> > >  Tue Oct 26 09:31:25 2021 [636647]: Espeak:
> > set_language_and_voice
> > > name=en+0
> > >  Tue Oct 26 09:31:25 2021 [636840]: Espeak: Successfully set
> > voice to
> > > "en+0"
> > >  Tue Oct 26 09:31:25 2021 [636850]: Espeak:
> > set_language_and_voice en 1
> > >  Tue Oct 26 09:31:25 2021 [636860]: Espeak:
> > set_language_and_voice
> > > name=en+0
> > >  Tue Oct 26 09:31:25 2021 [636973]: Espeak: Successfully set
> > voice to
> > > "en+0"
> > >  Tue Oct 26 09:31:25 2021 [687987]: Espeak: Rate set to 170.
> > >  Tue Oct 26 09:31:25 2021 [738788]: Espeak: Volume set to 200.
> > >  Tue Oct 26 09:31:25 2021 [937535]: Espeak: Pitch set to 50.
> > >  Tue Oct 26 09:31:25 2021 [987698]: Set punctuation mode.
> > >  Tue Oct 26 09:31:26 2021 [37917]: Set capitals mode.
> > >  Tue Oct 26 09:31:26 2021 [88426]: Espeak: Leaving
> > module_speak()
> > > normally.
> > >  Tue Oct 26 09:31:26 2021 [88689]: Espeak: Playback semaphore
> on.
> > >  Tue Oct 26 09:31:26 2021 [88729]: Printing reply: 701 BEGIN
> > >
> > > in the espeak.log but it's empty. Ah, may need to also set
> > espeak.conf
> > > to debug like this in /etc/speech-dispatcher/modules/espeak.conf:
> > >
> > > #Debugging
> > > Debug 1
> > >
> > > Change the 0 to 1 if it's there, otherwise add Debug 1 to the
> > end of
> > > the file if it's not there at all.
> > >
> > > thanks,
> > > Jeremy
> > >
> > > On Tue, Oct 26, 2021 at 10:53 AM Gustav Degreef
> > <gustav97 at gmail.com <mailto:gustav97 at gmail.com>
> > > <mailto:gustav97 at gmail.com <mailto:gustav97 at gmail.com>>> wrote:
> > >
> > >   Thanks Jeremy. Very clear instructions. Here is
> > the zip file,
> > >Â Â Â Gustav
> > >
> > >Â Â Â On 10/26/21 5:36 PM, Jeremy Whiting wrote:
> > >Â Â Â > Gustav,
> > >Â Â Â >
> > >Â Â Â > Ok, the quickest way to debug this will be to set a
> > LogDir in
> > >Â Â Â > /etc/speech-dispatcher/speechd.conf like this:
> > >Â Â Â >
> > >Â Â Â > #LogDir "default" <-- Comment out this one if it's not
> > by adding
> > >Â Â Â the #
> > >Â Â Â > at the front.
> > >Â Â Â > LogDir "/home/jeremy/sdlogs/"
> > >Â Â Â >
> > >Â Â Â > sudo will be needed to edit those files in
> > /etc/speech-dispatcher/
> > >Â Â Â >
> > >Â Â Â > you'll need to create an sdlogs folder in your home
> > directory (I
> > >Â Â Â tried
> > >Â Â Â > using the default opensuse /var/log/speech-dispatcher
> > path, but sd
> > >Â Â Â > running as me couldn't write there and it's
> > complicated, better to
> > >Â Â Â > just make a new path in your home folder and point to
> > it. Also use
> > >Â Â Â > your username, not jeremy :)
> > >Â Â Â >
> > >Â Â Â > Then when you run spd-say "Hello" or whatever it will
> > create a
> > >Â Â Â few log
> > >Â Â Â > files in that sdlogs path. I would say do that a couple
> > of times
> > >Â Â Â then
> > >Â Â Â > have okular speak the current page a couple of times.
> > Then zip
> > >Â Â Â up that
> > >Â Â Â > sdlogs folder and send it here and we can take a look
> > at what
> > >Â Â Â > speech-dispatcher is getting, and hopefully find out
> > why it sounds
> > >Â Â Â > garbled on your audio system.
> > >Â Â Â >
> > >Â Â Â > BR,
> > >Â Â Â > Jeremy
> > >Â Â Â >
> > >Â Â Â > On Tue, Oct 26, 2021 at 8:29 AM Jeremy Whiting
> > >Â Â Â <jpwhiting at kde.org <mailto:jpwhiting at kde.org>
> > <mailto:jpwhiting at kde.org <mailto:jpwhiting at kde.org>>
> > >Â Â Â > <mailto:jpwhiting at kde.org <mailto:jpwhiting at kde.org>
> > <mailto:jpwhiting at kde.org <mailto:jpwhiting at kde.org>>>> wrote:
> > >Â Â Â >
> > >Â Â Â >
> > >Â Â Â >
> > >   >   On Tue, Oct 26, 2021 at 8:09 AM Gustav
> Degreef
> > >Â Â Â <gustav97 at gmail.com <mailto:gustav97 at gmail.com>
> > <mailto:gustav97 at gmail.com <mailto:gustav97 at gmail.com>>
> > >   >   <mailto:gustav97 at gmail.com
> > <mailto:gustav97 at gmail.com> <mailto:gustav97 at gmail.com
> > <mailto:gustav97 at gmail.com>>>>
> > >Â Â Â wrote:
> > >Â Â Â >
> > >   >     Hi Jeremy,
> > >Â Â Â >
> > >   >     Deeply appreciate your time and
> > effort, but no luck.
> > >Â Â Â >
> > >   >   ÂÂ
> >  Â Â user:~/QtSpeech/hello_speak> qmake-qt5
> > >   >     Info: creating stash file
> > >   >   ÂÂ
> >  /home/user/QtSpeech/hello_speak/.qmake.stash
> > >   >     Project ERROR: Unknown module(s)
> > in QT: texttospeech
> > >Â Â Â >
> > >Â Â Â >
> > >   >   Ok, from that it seems you don't have the
> > -devel package
> > >Â Â Â needed to
> > >   >   build things against QtSpeech. For that
> > install the
> > >Â Â Â following:
> > >Â Â Â >
> > >   >   sudo zypper install libqt5-qtspeech-devel
> > >Â Â Â >
> > >   >   Then you should be able to build hello_speak.
> > >Â Â Â >
> > >   >   But since speech-dispatcher itself doesn't
> > have the problem
> > >   >   (spd-say didn't duplicate) the problem is
> > either in
> > >Â Â Â Okular's use
> > >   >   of QtSpeech (not likely, it's one line
> > that calls
> > >Â Â Â QtSpeech) or
> > >   >   QtSpeech itself. On my test vm I've only
> > got the speechd
> > >Â Â Â plugin
> > >   >   for QtSpeech in
> > >   >   /usr/lib64/qt5/plugins/texttospeech/
> > >   >   On your machine do you have a flite plugin
> > there also
> > >Â Â Â from some
> > >   >   package from 15.0 or 15.1 times maybe (I
> > don't see any flite
> > >   >   plugin package in 15.2 repos, but maybe
> > you got it before
> > >Â Â Â upgrading).
> > >Â Â Â >
> > >   >   I took another look at QtSpeech code
> > itself though and I
> > >Â Â Â don't see
> > >   >   any way it would ever use more than one
> > plugin or send
> > >Â Â Â the text to
> > >   >   multiple plugins. Then I took a look at
> > the speechd
> > >Â Â Â plugin again
> > >   >   and it definitely only sends the text it
> > gets to spd_say
> > >Â Â Â function
> > >   >   once. I think the next thing to try will
> > be to enable
> > >Â Â Â debugging in
> > >   >   speech-dispatcher (after we are sure
> > that's what plugin
> > >Â Â Â QtSpeech
> > >   >   is using) and see what text it's getting
> > to synthesize from
> > >   >   okular. I'll write up some instructions
> > for that once I
> > >Â Â Â do it here
> > >   >   and remember how it's done :)
> > >Â Â Â >
> > >   >   thanks,
> > >   >   Jeremy
> > >Â Â Â >
> > >   >     user:~/QtSpeech/hello_speak> make
> > >   >     make: *** No targets specified
> > and no makefile
> > >   found. Stop.
> > >Â Â Â >
> > >   >     Since your opensuse VM is
> > working fine, I doube
> > >Â Â Â checked my
> > >   >     installations. I have a
> > 17 inch laptop and a 13
> > >Â Â Â inch old one
> > >   >     that I use
> > >   >     for traveling. Both
> > machines have a main
> > >Â Â Â partition opensuse
> > >   >     15.2 and a
> > >   >     "test" partition with opensuse
> > 15.3. 3/4 of these
> > >   >     installations have
> > >   >     the duplicate voices in Okular
> > and Kmouth. Only 1 15.3
> > >   >     installation is
> > >   >     having a normal speech output in
> > Ojular. All 4
> > >Â Â Â installations
> > >   >     are
> > >   >     upgraded systems (from previous
> > opensuse 15.0, 15.1
> > >Â Â Â and prior
> > >   >     releases.Â
> > >   >     So none of them is a "clean"
> > install like your VM.
> > >Â Â Â >
> > >   >     I tried building it on each
> > laptop, but the same
> > >   result. ÂÂ
> > >   >     You were
> > >   >     reight at the beginnig, it's
> > turning out quite
> > >Â Â Â difficult to
> > >   >     track down.Â
> > >   >     Any other suggestions?
> > >Â Â Â >
> > >   >     If all else fails, I'll do a
> > clean install of 15.3
> > >Â Â Â on one of
> > >   >     these 4
> > >   >     partitions.
> > >Â Â Â >
> > >   >     Thanks, Gustav.
> > >Â Â Â >
> > >Â Â Â >
> > >   >   Truncating the rest to get past mailing
> > list size limit
> > >Â Â Â without
> > >   >   approval.
> > >Â Â Â >
> > >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-accessibility/attachments/20211027/8ccbbe5f/attachment-0001.htm>
More information about the kde-accessibility
mailing list