Text-to-speech: Difference between revisions

Content added Content deleted

Inline

Revision as of 00:16, 7 May 2023

Here are some ways to turn toki pona text into audio automatically. Unfortunately, they can't be used in accessibility tools, as none of them are integrated with Microsoft Speech API or similar. (If you want to change that, try making a language file for espeak-ng.)

word- and syllable-based by jan Sate (you can add your own voice, or use the sample one) - https://voice.sadale.net/
word-based by jan Ke Tami (with voices by jan Lakuse and kala Asi) - http://antetokipona.infinityfreeapp.com/kalama/
IPA input for other languages' TTS voices - http://ipa-reader.xyz/
neural model (Tacotron 2) by jan Kita:
- Uberduck - https://uberduck.ai/speak#mode=tts-basic&voice=hecko-tok
- Google Colab - link; model ID: 1-f-roUz_RQLvXm0bzpJYzO6_H_iDbFGJ; HiFi-GAN (optional): 1-0Py7GnQlbZxW6onijXPf0BFK98GBzAB

@@ Line 4: / Line 4: @@
 * word-based by jan Ke Tami (with voices by jan Lakuse and kala Asi) - http://antetokipona.infinityfreeapp.com/kalama/
 * IPA input for other languages' TTS voices - http://ipa-reader.xyz/
-* Tacotron 2 model by [[User:.hecko|jan Kita]]:
+* neural model (Tacotron 2) by [[User:.hecko|jan Kita]]:
 ** Uberduck - https://uberduck.ai/speak#mode=tts-basic&voice=hecko-tok
 ** Google Colab - [https://colab.research.google.com/drive/1wXWuhnw2pdfFy1L-pUzHfopW10W2GiJS link]; model ID: <code>1-f-roUz_RQLvXm0bzpJYzO6_H_iDbFGJ</code>; HiFi-GAN (optional): <code>1-0Py7GnQlbZxW6onijXPf0BFK98GBzAB</code>