Date   
Updates to Github #github

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

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

#729 formatted -X output in an easier way to parse by phonemizer

So the whole idea here is that we need to be able to map phonemized words back to their original text. When running -X with espeak, the original output looked like this:

tedi@Tedis-MacBook-Pro bin % ./espeak-ng  "this is a to be test" -X
Found: 'this' [DIs]  $u+ $strend $verbsf $nounf
Found: 'is' [Iz]   $pastf $only
Flags:  a   $nounf
Translate 'a'
  1	a        [a]
 26	_) a (_  [a#]

Found: 'to be
' [t@bi]   $pastf
Translate 'test'
  1	t        [t]

  1	e        [E]

  1	s        [s]

  1	t        [t]

 DIs Iz a# t@bi t'Est
tedi@Tedis-MacBook-Pro bin %

The new output will look like this:

tedi@Tedis-MacBook-Pro bin % ./espeak-ng "this is a to be test" -X
this~|||~DIs
is~|||~Iz
to be~|||~t@bi

The modified output is much easier to parse by the python wrapper contained inside Phonemizer because Phonemizer can now read in each line from the espeak-ng standard output, then split each line using the ~|||~ delimiter (e.g. string.split("~|||~") ). The text before the the delimiter is input text, while the text after the delimiter is ASCII IPA phonemized text, therefore the goal of being able to map phonemized text back to its original text has been achieved.

This is required for SSML processing inside Resembletron. The reason being that users apply tags to whole text words, the text words are then phonemized - if there is a 1 to 1 relationship between # of text words to # of phonemized words, then the tags can be translated easily. (E.g. hello, after phonemization becomes həlˈəʊ). Now assume the case where there is not a 1 to 1 relationship between # of text words to # of phonemised words, then a mapping of phonemized words to text words is required (which is now possible with this modification) For example, the words "to be" become one phonemized word "təbi" in certain sentences - with this word to phonemized word mapping, we can identify the original phonemized text.


[espeak-ng:master] New Comment on Pull Request #729 formatted -X output in an easier way to parse by phonemizer
By TediPapajorgji:

Sorry.... wrong repository...


[espeak-ng/espeak-ng] Pull request closed by TediPapajorgji:

#729 formatted -X output in an easier way to parse by phonemizer

So the whole idea here is that we need to be able to map phonemized words back to their original text. When running -X with espeak, the original output looked like this:

tedi@Tedis-MacBook-Pro bin % ./espeak-ng  "this is a to be test" -X
Found: 'this' [DIs]  $u+ $strend $verbsf $nounf
Found: 'is' [Iz]   $pastf $only
Flags:  a   $nounf
Translate 'a'
  1	a        [a]
 26	_) a (_  [a#]

Found: 'to be
' [t@bi]   $pastf
Translate 'test'
  1	t        [t]

  1	e        [E]

  1	s        [s]

  1	t        [t]

 DIs Iz a# t@bi t'Est
tedi@Tedis-MacBook-Pro bin %

The new output will look like this:

tedi@Tedis-MacBook-Pro bin % ./espeak-ng "this is a to be test" -X
this~|||~DIs
is~|||~Iz
to be~|||~t@bi

The modified output is much easier to parse by the python wrapper contained inside Phonemizer because Phonemizer can now read in each line from the espeak-ng standard output, then split each line using the ~|||~ delimiter (e.g. string.split("~|||~") ). The text before the the delimiter is input text, while the text after the delimiter is ASCII IPA phonemized text, therefore the goal of being able to map phonemized text back to its original text has been achieved.

This is required for SSML processing inside Resembletron. The reason being that users apply tags to whole text words, the text words are then phonemized - if there is a 1 to 1 relationship between # of text words to # of phonemized words, then the tags can be translated easily. (E.g. hello, after phonemization becomes həlˈəʊ). Now assume the case where there is not a 1 to 1 relationship between # of text words to # of phonemised words, then a mapping of phonemized words to text words is required (which is now possible with this modification) For example, the words "to be" become one phonemized word "təbi" in certain sentences - with this word to phonemized word mapping, we can identify the original phonemized text.


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

#729 formatted -X output in an easier way to parse by phonemizer

WRONG REPOSITORY SORRY!

So the whole idea here is that we need to be able to map phonemized words back to their original text. When running -X with espeak, the original output looked like this:

tedi@Tedis-MacBook-Pro bin % ./espeak-ng  "this is a to be test" -X
Found: 'this' [DIs]  $u+ $strend $verbsf $nounf
Found: 'is' [Iz]   $pastf $only
Flags:  a   $nounf
Translate 'a'
  1	a        [a]
 26	_) a (_  [a#]

Found: 'to be
' [t@bi]   $pastf
Translate 'test'
  1	t        [t]

  1	e        [E]

  1	s        [s]

  1	t        [t]

 DIs Iz a# t@bi t'Est
tedi@Tedis-MacBook-Pro bin %

The new output will look like this:

tedi@Tedis-MacBook-Pro bin % ./espeak-ng "this is a to be test" -X
this~|||~DIs
is~|||~Iz
to be~|||~t@bi

The modified output is much easier to parse by the python wrapper contained inside Phonemizer because Phonemizer can now read in each line from the espeak-ng standard output, then split each line using the ~|||~ delimiter (e.g. string.split("~|||~") ). The text before the the delimiter is input text, while the text after the delimiter is ASCII IPA phonemized text, therefore the goal of being able to map phonemized text back to its original text has been achieved.

This is required for SSML processing inside Resembletron. The reason being that users apply tags to whole text words, the text words are then phonemized - if there is a 1 to 1 relationship between # of text words to # of phonemized words, then the tags can be translated easily. (E.g. hello, after phonemization becomes həlˈəʊ). Now assume the case where there is not a 1 to 1 relationship between # of text words to # of phonemised words, then a mapping of phonemized words to text words is required (which is now possible with this modification) For example, the words "to be" become one phonemized word "təbi" in certain sentences - with this word to phonemized word mapping, we can identify the original phonemized text.

[espeak-ng:master] reported: No MBROLA support for Mac #github

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

[espeak-ng:master] New Comment on Issue #336 No MBROLA support for Mac
By gogobd:

This issue still exists in espeak: stable 1.48.04 (bottled) installed via brew on MacOS High Sierra

[espeak-ng:master] new issue: issues with russian language #github

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

[espeak-ng:master] New Issue Created by beqabeqa473:
#728 issues with russian language

hello all.

there are some issues with russian language in espeak-ng which would be good to fix.

  1. numbers are not working as they should. e.g. 1000000, 1000000000, 1000000000000

  2. letter combination of ьи is pronounced incorrectly e.g. in words статьи, воробьи, ручьи, last letter я is not correct/ it should be и instead.

thanks.

Re: editing klatt, and segmenting phonemes.

David Hilton
 

so, there is a version of espeakedit for espeak ng, good to know. What is the jedit java editor used for, the phoneme separation, or the klatt synthesis.

On Thu, Apr 2, 2020 at 12:13 AM Valdis Vitolins via groups.io <valdis.vitolins=odo.lv@groups.io> wrote:
https://github.com/valdisvi/espeak-ng-espeakedit
or
https://github.com/valdisvi/espeak-ng-jeditor

Valdis

> Hello, I was wondering how would I go about editing the klatt
> synthesizer in espeak. Also is there a version of espeakedit for
> espeak ng? If not, how would I go about segmenting existing phonemes,
> to create new phomenes. thank you for your time. david hilton davidhi
> lton936@...




[espeak-ng:master] reported: Typing interrupt doesn't work when using espeakup with espeak-ng #github

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

[espeak-ng:master] New Comment on Issue #285 Typing interrupt doesn't work when using espeakup with espeak-ng
By valdisvi:

@am-dxer, then probably this issue can be solved by switching to the latest stable version of eSpeak NG.

Re: editing klatt, and segmenting phonemes.

Valdis Vitolins
 

Hello, I was wondering how would I go about editing the klatt
synthesizer in espeak. Also is there a version of espeakedit for
espeak ng? If not, how would I go about segmenting existing phonemes,
to create new phomenes. thank you for your time. david hilton davidhi
lton936@...

editing klatt, and segmenting phonemes.

David Hilton
 

Hello, I was wondering how would I go about editing the klatt synthesizer in espeak. Also is there a version of espeakedit for espeak ng? If not, how would I go about segmenting existing phonemes, to create new phomenes. thank you for your time. david hilton davidhilton936@...

[espeak-ng:master] reported: Typing interrupt doesn't work when using espeakup with espeak-ng #github

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

[espeak-ng:master] New Comment on Issue #285 Typing interrupt doesn't work when using espeakup with espeak-ng
By alex19EP:

@valdisvi 1.5x

[espeak-ng:master] reported: Typing interrupt doesn't work when using espeakup with espeak-ng #github

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

[espeak-ng:master] New Comment on Issue #285 Typing interrupt doesn't work when using espeakup with espeak-ng
By valdisvi:

Which version are you using — 1.4x or 1.5x?

Updates to Github #github

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

[espeak-ng:master] New Comment on Issue #463 still overlapping speech with espeak-ng
By alex19EP:

can not confirm this on current espeak-ng and espeakup


[espeak-ng:master] New Comment on Issue #285 Typing interrupt doesn't work when using espeakup with espeak-ng
By alex19EP:

can not confirm on current espeakup and espeak-ng

[espeak-ng:master] new issue: Klatt synthesizer has issues with Japanese language #github

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

[espeak-ng:master] New Issue Created by The1Freeman:
#727 Klatt synthesizer has issues with Japanese language

For the record: I am no expert in the japanese language and am thus tinkering around with this, with that said however, the Japanese language in eSpeak-NG combined any of the klatt voices (and the announcer voice) will produce weirdly sounding artifacts on certain phonemes, particularly with [[v]], [[n]], [[N]] and [[m]].

The best way to describe the glitch in the klatt synthesizer would be that it is too nasal with those phonemes whereas in the default voice it is not the case, the rest seem fine though, however with [[N]] the higher formants are loud, when that is not the case with the default voice.

klatt_glitch.zip

Attached is a zip archive containing two wav files which showcase this glitch in the japanese language, generated with [[vv]]. [[nn]]. [[NN]]. [[m]].

Updates to Github #github

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

[espeak-ng:master] New Issue Created by spresse1:
#725 ' (quote) versus " (quotes) is almost impossible to audibly differentiate

Users of screen readers (eg: Orca, NVDA, JAWS) and other TTS assistive technologies rely on accurate and easy-to-understand reading of the text. One context in which this becomes especially critical is systems administration and software development, where TTS users need to be able to clearly and easily differentiate types of punctuation with minimal effort in the midst of reading a larger portion of text. For example, the two characters have significantly different meanings in shells and in some programming languages.

In English, eSpeak-ng currently reads the single quote character (') as "quote" and the double quote character (") as "quotes". These two words are very difficult to differentiate, especially at the higher reading speeds often used by TTS users.

eSpeak-ng should read these two characters in a manner which makes them easy to differentiate.

There are three possibilities which come to mind:

  1. Read the single quote as "tick". This option keeps everything short and has a nice symmetry with the backtick (`) character. This behavior would also be consistent with other synthesizers, such as Nuance's Vocalizer and Windows OneCore Voices. However, "tick" is a colloquial name for this character and sometimes also refers to a check mark.
  2. Read the single quote as "apostrophe". This option is mostly correct and will create the requested reading clarity. However, "apostrophe" is a relatively long word and will slow down the reading. But apostrophe indicates a use of the single-quote character and is not the name of the character. This option would be consistent with Apple's VoiceOver.
  3. Read the double quote as "double quotes". Again, this option is longer than current ways of reading the characters, but will result in a correct reading of this punctuation under all circumstances. I'm not aware of any synthesizer or screen reader which uses this option.

I have spoken with one TTS assistive technology user who would prefer option 1. However, I don't have the depth of knowledge required to make this choice for all eSpeak-ng users. I'm happy to do all the work for this change (and will file a PR for option 1 shortly), but am opening this issue such that those with more experience can direct me on what the best version of this change is.


[espeak-ng/espeak-ng] Pull request opened by spresse1:

#726 #725: ' (quote) versus " (quotes) is almost impossible to audibly differentiate

Users of screen readers (eg: Orca, NVDA, JAWS) and other TTS assistive technologies rely on accurate and easy-to-understand reading of the text. One context in which this becomes especially critical is systems administration and software development, where TTS users need to be able to clearly and easily differentiate types of punctuation with minimal effort in the midst of reading a larger portion of text. For example, the two characters have significantly different meanings in shells and in some programming languages.

This PR changes the name of the quote (') character to "tick", as this was difficult to differentiate from """, which reads as "quotes"


[espeak-ng:master] New Comment on Issue #725 ' (quote) versus " (quotes) is almost impossible to audibly differentiate
By spresse1:

I opened PR#726 as an example of implementing option 1.

[espeak-ng:master] new issue: ' (quote) versus " (quotes) is almost impossible to audibly differentiate #github

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

[espeak-ng:master] New Issue Created by spresse1:
#725 ' (quote) versus " (quotes) is almost impossible to audibly differentiate

Users of screen readers (eg: Orca, NVDA, JAWS) and other TTS assistive technologies rely on accurate and easy-to-understand reading of the text. One context in which this becomes especially critical is systems administration and software development, where TTS users need to be able to clearly and easily differentiate types of punctuation with minimal effort in the midst of reading a larger portion of text. For example, the two characters have significantly different meanings in shells and in some programming languages.

In English, eSpeak currently reads the single quote character (') as "quote" and the double quote character (") as "quotes". These two words are very difficult to differentiate, especially at the higher reading speeds often used by TTS users.

eSpeak-ng should read these two characters in a manner which makes them easy to differentiate.

There are three possibilities which come to mind:

  1. Read the single quote as "tick". This option keeps everything short and has a nice symmetry with the backtick (`) character. This behavior would also be consistent with other synthesizers, such as Nuance's Vocalizer and Windows OneCore Voices. However, "tick" is a colloquial name for this character and sometimes also refers to a check mark.
  2. Read the single quote as "apostrophe". This option is mostly correct and will create the requested reading clarity. However, "apostrophe" is a relatively long word and will slow down the reading. But apostrophe indicates a use of the single-quote character and is not the name of the character. This option would be consistent with Apple's VoiceOver.
  3. Read the double quote as "double quotes". Again, this option is longer than current ways of reading the characters, but will result in a correct reading of this punctuation under all circumstances. I'm not aware of any synthesizer or screen reader which uses this option.

I have spoken with one TTS assistive technology user who would prefer option 1. However, I don't have the depth of knowledge required to make this choice for all eSpeak-ng users. I'm happy to do all the work for this change (and will file a PR for option 1 shortly), but am opening this issue such that those with more experience can direct me on what the best version of this change is.

[espeak-ng:master] reported: Does this codecs compatible with espeakNG ? #github

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

[espeak-ng:master] New Comment on Issue #724 Does this codecs compatible with espeakNG ?
By valdisvi:

eSpeak NG is designed as drop-in replacement for eSpeak. Therefore it should work.

[espeak-ng:master] reported: Does this codecs compatible with espeakNG ? #github

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

[espeak-ng:master] New Comment on Issue #724 Does this codecs compatible with espeakNG ?
By The1Freeman:

I would test it out and see what happens, it's sort of pointless by asking this question without having it tested yourself.

I'd test it, and if it doesn't work, open a issue with detailed descriptions and logs when possible.

[espeak-ng:master] new issue: Does this codecs compatible with espeakNG ? #github

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

[espeak-ng:master] New Issue Created by Nokia808:
#724 Does this codecs compatible with espeakNG ?

Hi. I'm on Fedora Linux 30 X64 bit Cinnamon edition. On Fedora Linux the default TTS is espeakNG & it is the only one that existing on my system & no other program.

There is a codec in GNOME software center for espeak - see screenshot bellow:

espeak-codecs

My question is that: is this codecs compatible with espeakNG also or not & being only usable with old espeak program ?

Thanks.

Re: ESPEAK 101 AND MORE LANGUAGES?

Simon Eigeldinger
 

Hi ricky,


Get espeak-ng 1.50.
Though because of the installer currently not including all new languages you might have not all that many more.

Here are the links to the files.

64 bit:
https://github.com/espeak-ng/espeak-ng/releases/download/1.50/espeak-ng-20191129-b702b03-x64.msi

32 bit:
https://github.com/espeak-ng/espeak-ng/releases/download/1.50/espeak-ng-20191129-b702b03-x86.msi

Greetings,
Simon



Am 28.03.2020 um 20:38 schrieb Ricky Lomey:

Hi
The version of Espeak I'm using, been since 2018, has seventy-nine options
but how can I update to the latest, I think there are now more than 101
languages?
Thanks.
Ricky Lomey

ESPEAK 101 AND MORE LANGUAGES?

Ricky Lomey
 

Hi

The version of Espeak I'm using, been since 2018, has seventy-nine options
but how can I update to the latest, I think there are now more than 101
languages?

Thanks.

Ricky Lomey

[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 mercster:

Hey folks... in Ubuntu i have a zillion files in /etc/speech-dispatcher and downwards... I can't tell how the default voice is getting set. I want to change it to "robert". The default voice on my system sounds like MALE3/en-d , but I've tried replacing "en-d" for "robert" in the espeak-generic.conf and it's not changing the default. Can anyone help me?

Thanks.

Github push to espeak-ng:espeak-ng #github

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

1 New Commit:

[espeak-ng:master] By Ejaz Shah <eashah67@...>:
c4a9f7f52c31: Improvements for Urdu

Modified: dictsource/ur_list
Modified: dictsource/ur_rules