Date   

[espeak-ng:master] reported: How to change default voice #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By OakLD:

My mistake, I forgot to mention speech-dispatcher-module-espeak 0.8.8-lp151.3.6.1. It provides:

/etc/speech-dispatcher/modules/espeak.conf
/usr/lib64/speech-dispatcher-modules/sd_espeak

You're right that there's no espeak-ng module for speech-dispatcher available in OpenSUSE repositories. MBROLA is available, but nly German voices (!) are supplied. However, I installed some from Fedora. They're noarch, so it should be fine. There are also some packages I dind't install (didn't think they're needed), such as espeak-ng-compat and espeakedit. It's really pitty that I got my festival working with relatively quality HTS voices, I got my espeak working with MBROLA (yes, they're working when forced via a parameter in command line), but neither I can get it working with speech-dispatcher... It's so desperate that I'm looking into option to write my own plugin. But it's not a small job...


Updates to Github #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By DidierSpaier:

Sorry then I'm afraid you are out of luck with OpenSuse. Here is why.

I had a look at the packages you mention from the official repository (I ma not running this distribution) and unless I miss something - espeak and espeak-ng do not include the Klatt variants like Robert (and you don't have the espeak-ng module of speech-dispatcher) - there is no packages for the mbrola voices, only the engine. - there is no package for the svox-pico voices - maybe you could get a voxin synthesizer from Oralux and associated voices, but I don't know if they have packages for OpenSuse (and the voxin website is off-line for maintenance at the moment so I can't check now).

So, the only way to get decent voices as far as I know is to use another Linux distribution, for instance Debian, or Slint (that I maintain), or Arch.


[espeak-ng:master] New Comment on Issue #719 How to change default voice
By DidierSpaier:

Sorry then I'm afraid you are out of luck with OpenSuse. Here is why.

I had a look at the packages you mention from the official repository (I am not running this distribution) and unless I miss something: - espeak and espeak-ng do not include the Klatt variants like Robert (and you don't have the espeak-ng module of speech-dispatcher) - there is no packages for the mbrola voices, only the engine. - there is no package for the svox-pico voices - maybe you could get a voxin synthesizer from Oralux and associated voices, but I don't know if they have packages for OpenSuse (and the voxin website is off-line for maintenance at the moment so I can't check now).

So, the only way to get decent voices as far as I know is to use another Linux distribution, for instance Debian, or Slint (that I maintain), or Arch.


Updates to Github #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By OakLD:

Oh, it's not secret. I have OpenSUSE Leap 15.1 Linux, espeak 1.48.04-lp151.2.3, espeak-ng 1.49.2-lp151.4.3, speech-dispatcher 0.8.8-lp151.3.6.1. The lp151 indicates all of those are default Leap 15.1 packages from the official OpenSUSE repositories. The speechd.conf in short: LogLevel 3 LogDir "default" DefaultVolume 100 AddModule "espeak" "sd_espeak" "espeak.conf"


[espeak-ng:master] New Comment on Issue #719 How to change default voice
By OakLD:

Oh, it's not secret. I have OpenSUSE Leap 15.1 Linux, espeak 1.48.04-lp151.2.3, espeak-ng 1.49.2-lp151.4.3, speech-dispatcher 0.8.8-lp151.3.6.1. The lp151 indicates all of those are default Leap 15.1 packages from the official OpenSUSE repositories. The speechd.conf in short:

LogLevel  3
LogDir  "default"
DefaultVolume 100
AddModule "espeak"       "sd_espeak"   "espeak.conf"
AddModule "dummy"         "sd_dummy"      ""
DefaultModule espeak
Include "clients/*.conf"

The software is XP11 simulator with ATC 124th plugin.


[espeak-ng:master] reported: I'm using espeak to generate pronunciations, for a list of words sometimes there will be no space in between two of them #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

Is there something one can do to make sure there always is a space?


Updates to Github #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #739 I'm using espeak to generate pronunciations, for a list of words sometimes there will be no space in between two of them
By valdisvi:

It is because in pronunciation rules, where only secondary stress is put without word boundary.


[espeak-ng:master] New Comment on Issue #739 I'm using espeak to generate pronunciations, for a list of words sometimes there will be no space in between two of them
By valdisvi:

It is because in pronunciation rules only secondary stress is put without word boundary.


[espeak-ng:master] new issue: I'm using espeak to generate pronunciations, for a list of words sometimes there will be no space in between two of them #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Issue Created by RuABraun:
#739 I'm using espeak to generate pronunciations, for a list of words sometimes there will be no space in between two of them

I'm calling espeak-ng with -v en-us -f fin -q -x --ipa=1 --phonout=fout. And most of the time I can have a list of words and the file, and I'll get pronunciations separated by a space.

However for some words the space is missing, so for two words I have one pronunciation, and I don't know how to split them up.

One example is the words "there are", which results in a sequence phones with no space inbetween.

Is there something one can do to make sure there always is a space?


Updates to Github #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By DidierSpaier:

Then I can probably help, but only if you tell exactly which speech-dispatcher version you have installed and which synthesizers and voices you have. Still the exact version of your distribution and version will give clues. Also it would be useful to know your /etc/speech-dispatcher.speechd.conf file and the content of /etc/speech-dispatcher/modules. Fell free to send me this information privately @ didier at slint dot fr. Out of curiosity, which is this software you bought? I am interested as the maintainer of an accessible Linux distribution to know what is available.

Cheers,

Didier


[espeak-ng:master] New Comment on Issue #719 How to change default voice
By DidierSpaier:

I meant /etc/speech-dispatcher/speechd.conf


[espeak-ng:master] reported: How to change default voice #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By OakLD:

No I don't have Orca installed and unfortunately it is not related to my issue and can't help me. I purchased a software and for that software a plugin (I had a choice of 1 for Linux, unfortunately), that works with speech-dispatcher only. So I have free hands for what stands beyond speech-dispatcher, but I can't replace it. The default espeak voice is so horrible it's not really understandable. So I'm simply after better TTS voice quality, but I don't need perfect, just not that horrible. I'd say MBROLA is minimum acceptable quality.

I'm quite confused, whether I can set a default voice in espeak and also whether I can set them for espeak in speach-dispatcher (I know this 2nd part is a bit off topic here, I didn't ask for it originally, but since it was brought out...).


Updates to Github #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By DidierSpaier:

n@OakLD: Do you use Orca? If yes, and you have recent versions of speech-dispatcher and Orca, in the Preferences GUI of Orca in the Voices Tab you can choose the synthesizer, then the language, then the voice. Obviously the choices depend on which synthesizers and voices you have installed

For instance here (Slint64-14.2.1..2) if I choose as synthesizer espeak-ng and as language en-US i can choice as voice (labeled Person) English (America) + Robert.

If I choose as synthesizer espeak-ng-mbrola-generic and as language en-us I can choose among us1, us2 or us3 (that map respectively to us1, us2 or us3).

To get a good voice you could also get the voxin synthesizer and associated voices from oralux.org.

To know what's available to you, please indicate which distribution and version you are using, and if not Orca which screen reader. Is it speechd-up or fenrir?

If not using Orca, setting a specific voice for speech-dispatcher is currently tricky, because it is only possible editing a config file to map this voice whit a "generic" one line MALE3.

I could file an issue about that against speech-dispatcher but I am pretty sure that Samuel will respond "patches are welcome" and I am currently busy .

As an aside, the issue you filed here would better fit in the speech-dispatcher mailing list anyway, as it is not specific to espak-ng.

Cheers,

Didier


[espeak-ng:master] New Comment on Issue #719 How to change default voice
By DidierSpaier:

Instead of: If I choose as synthesizer espeak-ng-mbrola-generic and as language en-us I can choose among us1, us2 or us3 (that map respectively to us1, us2 or us3).

Please read: If I choose as synthesizer espeak-ng-mbrola-generic and as language en-us I can choose among us1, us2 or us3 (that map respectively to mb-us1, mb-us2 or mb-us3).


[espeak-ng:master] reported: How to change default voice #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By OakLD:

@DidierSpaier I don't think so. This only defines what engine and what setting to set the dispatcher itself to. But there are files in ~/.config/speech-dispatcher/modules/ for each engine, i.e. festival.conf, or in this case espeak.conf, to set engine-specific stuff. I believe the voices should be set here. I think it should be here you set i.e. MALE1 to mb-us1 Unfortunately, this file does not provide any example to set that and I failed to google anything on this subject.


[espeak-ng:master] reported: How to change default voice #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #719 How to change default voice
By OakLD:

@valdisvi I wonder, how do you set speech-dispatcher to use MBROLA engine directly (i.e. not via espeak or festival). It seems that is what you suggest and should be possible. I have it installed, but it isn't mentioned in the speech-dispatcher configuration file (like many other engines). When I try a wild guess, it doesn't work:

 #AddModule "espeak"       "sd_espeak"   "espeak.conf"
 #AddModule "espeak-ng"    "sd_espeak-ng" "espeak-ng.conf"
 AddModule "mbrola"    "sd_mbrola" "mbrola.conf"
 #AddModule "festival"     "sd_festival"  "festival.conf"
 #AddModule "flite"        "sd_flite"     "flite.conf"

I couldn't google anything either...


[espeak-ng:master] reported: Abbreviations followed by period sound different #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #735 Abbreviations followed by period sound different
By tspivey:

I can hear the difference on that page. It's slight, but noticeable. Listen specifically to the part after mister and how it flows into the next word.


[espeak-ng:master] reported: Abbreviations followed by period sound different #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Issue #735 Abbreviations followed by period sound different
By valdisvi:

There is difference in line breaks for traced output, but otherwise I don't see or hear any difference.


Pull Request Updated #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng/espeak-ng] Pull request updated by rongcuid:

#738 Improves on some Chinese pronunciation

I improved part of the the Chinese dictionaries and their priorities, fixing some common unclear/wrong pronunciations. These are not comprehensive, just some I discovered when using espeak-ng for my project.


Pull Request Opened #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng/espeak-ng] Pull request opened by rongcuid:

#738 Improves on some Chinese pronunciation

I improved part of the the Chinese dictionaries and their priorities, fixing some common unclear/wrong pronunciations. These are not comprehensive, just some I discovered when using espeak-ng for my project.


[espeak-ng:master] new issue: SSML input <voice name="value" age="value"> changes to default voice mid-sentence #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Issue Created by guest271314:
#737 SSML input changes to default voice mid-sentence

Run the tests at https://github.com/brailcom/speechd/blob/bdf6557ac0fea1ba5efbcffc1284e5f2784371d0/src/tests/ssml.test at command line or from files.

For input https://github.com/brailcom/speechd/blob/bdf6557ac0fea1ba5efbcffc1284e5f2784371d0/src/tests/ssml.test#L95 (input.txt)

<speak>
<voice age="8">Human beings are the only creatures that allow their children
to come back home.</voice>
        -- Bill Cosby (1937 - )
</speak>

$ espeak-ng -m -f input.txt

the voice changes from age="8" to the default voice between "al" and "low" at "allow". When name="Storm" is substituted for age="8" the voice also changes to default voice at the same syllable.

When name="gmw/en-029" is set the voice does not change mid-sentence which is the expected result.


Killing mispronounced

Travis Siegel
 

Not sure how to make it happen again, but yesterday when I was reading a book, I found that espeak pronounced killing as if it rhymed with trailing.
It was written like:
"Killing
I don't know if it was the app (kindle), or the character set, but it definitely confused me.
For what it's worth, it isn't doing it now using the dos prompt and ssh to my linux box to write this.
Any clues on this one?
I use the ivan voice at 99% speed, (100 percent breaks up too much).


[espeak-ng:master] new issue: How to set SSML on by default for /emscripten directory? #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Issue Created by guest271314:
#736 How to set SSML on by default for /emscripten directory?

https://github.com/espeak-ng/espeak-ng/tree/master/emscripten does not set SSML option (-m) to on by default nor provide a means to do so.

Adding a <break> element to demo https://www.readbeyond.it/espeakng/ results in "break" then "time" being output to speakers.

How to update code in /emscripten directory to set SSML parsing on by default for all synthesis?


[espeak-ng:master] new issue: Abbreviations followed by period sound different #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Issue Created by tspivey:
#735 Abbreviations followed by period sound different

Testing with en (and en-us) latest master 1764ec647f19258a19665e227c6d15850139963c.

  1. Mr. Smith
  2. Mr Smith

I think 1 should sound like 2, since the abbreviation is just replacing the word.


Updates to Github #github

espeak-ng@groups.io Integration <espeak-ng@...>
 

2 New Commits:

[espeak-ng:master] By Valdis Vitolins <valdis.vitolins@...>:
c8512c305f47: Reorder makefile entries

Modified: Makefile.am


[espeak-ng:master] By Valdis Vitolins <valdis.vitolins@...>:
1764ec647f19: tr: issue #734: update Turkish language

Added: dictsource/tr_listx
Modified: CHANGELOG.md
Modified: dictsource/tr_list


[espeak-ng:master] New Comment on Issue #734 Update for Turkish
By valdisvi:

Thanks for your contribution! It is added in project.


[espeak-ng:master] New Comment on Issue #733 [Help Needed] Is there any docs for implementing a custom language?
By valdisvi:

Look at Adding or Improving a Language.

901 - 920 of 3963