Skip to main content


A demo of a #spiel sample app, the voices used in order are: eSpeakNG's "Andy" variant, MBROLA US2, and Piper's Amy. You can observe the different features like word tracking and quality. #speech #tts #gnome #linux

Federico Mena Quintero reshared this.

in reply to Eitan

That espeak-ng voice sounds awful. And I routinely use espeak-ng. But I use the Max variant, which is the default in NVDA.
in reply to Eitan

Wow, Piper sounds much better, and I don’t even speak a word Arabic. Piper is the second one, right?
in reply to Marco Zehe

@Marco Yup. A lot of work went into that with a new diacritization library for proper vowels. I hope to one day make a similar library for Hebrew diacritization.
in reply to Eitan

wow. the second one is so much clearer. that's really a huge improvement
in reply to Eitan

Great job @eeejay 👏. Even if i don't speak a word Arabic the Piper TTS improvement can absolutely be heard.

Btw. @Marco if you are interested in Piper usage or clone your voice for #tts maybe one or two of my video tutorials are helpful:
https://youtu.be/b_we_jma220
or https://youtu.be/GGvdq3giiTQ

in reply to Thorsten-Voice

@thorstenvoice @Marco I haven't directly worked on Piper, this is other folks' hard work. I am just integrating it into a new desktop speech framework, #spiel.