0

The TTS pronounces "az-zumaru" instead of "az-zumar" when passed the following Arabic word (title of a Sura): ٱلزُّمَر

Any suggestions what to do to produce the correct speech? Is there some unicode character that will tell the TTS engine that "no Tanween is needed here" (for example)? My guess is it is adding a tanween when it should not, but I'm not sure as I'm not an expert in this area.

I've tried different spellings (without diacritics) from https://en.wikipedia.org/wiki/Az-Zumar but it seems to pronounce it with a trailing "u" sound in all cases.

If part of a longer sentence, the word appears to be pronounced correctly.

I've also tried different android voices (e.g., ar-xa-x-arz-local, ar-xa-x-arc-local, etc.) and they all seem to add a trailing sound for single words.

F C
  • 143
  • 1
  • 7

2 Answers2

1

If you generate your text to speech with SSML you have the phoneme element.

With this commando you can produce any sound from your language without any limitation. Some knowledge of phonemes is required. Here is an example of how you can pronounce tomato with the phoneme and the ipa alphabet

Follow this link for more information

https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/speech-synthesis-markup-pronunciation

<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="en-US">
    <voice name="en-US-JennyNeural">
        <phoneme alphabet="ipa" ph="tə.ˈmeɪ.toʊ"> tomato </phoneme>
    </voice>
</speak>
Peter Adriaenssens
  • 72
  • 2
  • Thank you. Would you have an example of how SSML can be used with Android TTS? Here's what I found and it wasn't very clear https://stackoverflow.com/questions/62436229/ssml-support-in-android-tts – F C Apr 24 '23 at 08:43
0

What worked for me was to add a unicode sukun character at the end of each word: ْ

Source: A friend who has done work in the area.

See also: https://en.wiktionary.org/wiki/sukun

F C
  • 143
  • 1
  • 7