New Urdu voice: Fasih (ur_PK male, medium quality)

#70
by IhorShevchuk - opened

New Urdu Voice: Fasih

I've trained a new Urdu (ur_PK) male voice for Piper called Fasih.

Details

  • Language: Urdu (ur_PK)
  • Gender: Male
  • Quality: Medium
  • Sample Rate: 22050 Hz
  • Phonemizer: eSpeak (ur)

Download

The model is available here:
https://huggingface.co/IhorShevchuk/piper-voice-ur-fasih

Motivation

Urdu is still missing from many mainstream TTS systems, including Apple’s built-in voices.

I wanted to help fill that gap and make something useful for Urdu speakers - especially for accessibility use cases like screen readers or apps(like Piper Apple app) for people who rely on voice output.

Rhasspy org

Thanks @IhorShevchuk ! I've merged the voice here and generated a sample over on the Piper samples page: https://rhasspy.github.io/piper-samples/
Thank you for your contribution 🙂

My dear brother, I have a sincere request: Please release an Urdu Speech-to-Text model specifically optimized to run smoothly on a CPU. To date, there isn't a single high-quality model available that works efficiently on standard personal computers; most existing ones struggle to transcribe Urdu accurately and often produce incorrect results.
​Furthermore, while your Urdu TTS (Text-to-Speech) is a good start, it needs more depth. If you could integrate human emotions and natural expressions into it, the voice would feel much more lifelike and impactful. Please, prioritize a fast and accurate solution for PC users."

please please please please 🥺🥺🥺

huggingface:-

https://huggingface.co/IhorShevchuk/piper-voice-ur-fasih

My dear brother, I have a sincere request: Please release an Urdu Speech-to-Text model specifically optimized to run smoothly on a CPU. To date, there isn't a single high-quality model available that works efficiently on standard personal computers; most existing ones struggle to transcribe Urdu accurately and often produce incorrect results.
​Furthermore, while your Urdu TTS (Text-to-Speech) is a good start, it needs more depth. If you could integrate human emotions and natural expressions into it, the voice would feel much more lifelike and impactful. Please, prioritize a fast and accurate solution for PC users."

Sign up or log in to comment