Topics

[espeak-ng:master] new issue: eSpeak-NG produces slurred speech under certain conditions #github

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

[espeak-ng:master] New Issue Created by The1Freeman:
#671 eSpeak-NG produces slurred speech under certain conditions

eSpeak-NG will produce slurred speech when placing a [[_^_JA custom pronounciation]] tag somewhere in a text. (I don't know if this affects other languages, so this is the first instance in where it produces such weird output).

This apparently happens when the sentence contains a comma, and somewhere in the next part of the text after the comma where that tag is set, it'll switch to a (presumably) different language while retaining the original voice (in this case I assume it uses the japanese phoneme set, which does not contain all of the phonemes english has).

Here's the text that's causing this problem: from wikipedia: According to director Imaishi, much of the plot is based on his observation that the Japanese manner of pronouncing "fascio" ( [[_^_JA p\as\o:]] fassho) is nearly the same as the word "fashion" ( [[_^_JA p\as\o:n]] fasshon), his observation that the pronunciation of the Japanese words "school uniform" ( [[_^_JA se:if_'ku]] seifuku) and "conquest" ( [[_^_JA se:if_'ku]] seifuku) are identical, and that the titular kiru may mean "kill" ( [[_^_JA k'iru]]), "to cut" ( [[_^_JA k'iru]]), or "to wear" ( [[_^_JA k'iru]]).`

You can work this around by placing [[_^_EN-US ]] after the comma, in which eSpeak-NG will read the text proper.

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

[espeak-ng:master] New Issue Created by The1Freeman:
#671 eSpeak-NG produces slurred speech under certain conditions

eSpeak-NG will produce slurred speech when placing a [[_^_JA custom pronounciation]] tag somewhere in a text. (I don't know if this affects other languages, so this is the first instance in where it produces such weird output).

This apparently happens when the sentence contains a comma, and somewhere in the next part of the text after the comma where that tag is set, it'll switch to a (presumably) different language while retaining the original voice (in this case I assume it uses the japanese phoneme set, which does not contain all of the phonemes english has).

Here's the text that's causing this problem: from wikipedia: According to director Imaishi, much of the plot is based on his observation that the Japanese manner of pronouncing "fascio" ( [[_^_JA p\as\o:]] fassho) is nearly the same as the word "fashion" ( [[_^_JA p\as\o:n]] fasshon), his observation that the pronunciation of the Japanese words "school uniform" ( [[_^_JA se:if_'ku]] seifuku) and "conquest" ( [[_^_JA se:if_'ku]] seifuku) are identical, and that the titular kiru may mean "kill" ( [[_^_JA k'iru]]), "to cut" ( [[_^_JA k'iru]]), or "to wear" ( [[_^_JA k'iru]]).`

You can work this around by placing [[_^_EN-US ]] after the comma, in which eSpeak-NG will read the text proper.

Edit: This happens with other languages. Switch the language code to any other from the tag and eSpeak will attempt to read it in said language (which will fail) Example text: A test for eSpeak, after the comma the text should be read in a weird way ( [[_^_DE p\as\o:]] fassho).