CoolFace
Modelpublic

tinisoft/piper-ta_IN-rasa_female-medium

sourceHugging Facecc-by-4.0updated 4mo agoView on Hugging Face
0likes248downloads
Model Card

Piper Tamil — Rasa Female (medium)

A Piper text-to-speech voice for Tamil (ta_IN), trained on the female speaker of the AI4Bharat Rasa expressive speech dataset.

  • Language: Tamil (ta_IN)
  • Sample rate: 22050 Hz
  • Quality: medium
  • Phoneme backend: eSpeak (voice: ta)
  • Framework: Piper (ONNX)

Files

FilePurpose
ta_IN-rasa_female-medium.onnxVoice model weights
ta_IN-rasa_female-medium.onnx.jsonVoice config (phoneme map, inference params)
sample.wavDemo synthesis (see "Sample sentence" below)

Usage

Install Piper:

bash
pip install piper-tts

Synthesize:

bash
echo "வணக்கம், இது தமிழ் குரல் மாதிரி." \
  | piper -m ta_IN-rasa_female-medium.onnx -f out.wav

Or from Python:

python
from piper import PiperVoice
import wave

voice = PiperVoice.load("ta_IN-rasa_female-medium.onnx")
with wave.open("out.wav", "wb") as wav:
    voice.synthesize("வணக்கம், இது தமிழ் குரல் மாதிரி.", wav)

Sample sentence

வணக்கம், இது பைப்பர் தமிழ் மாதிரிகளின் ஒப்பீட்டு சோதனை. இந்த வாக்கியம் ஒவ்வொரு குரல் மாதிரியிலும் ஒரே மாதிரியாக உருவாக்கப்படுகிறது.

License

Released under CC-BY-4.0, matching the license of the underlying Rasa dataset.

Attribution

This model is trained on the AI4Bharat Rasa dataset. If you use this voice, please cite:

bibtex
@inproceedings{ai4bharat2024rasa,
  author    = {Praveen Srinivasa Varadhan and Ashwin Sankar and Giri Raju and Mitesh M. Khapra},
  title     = {{Rasa: Building Expressive Speech Synthesis Systems for Indian Languages in Low-resource Settings}},
  booktitle = {Proc. INTERSPEECH 2024},
  year      = {2024}
}

Dataset: ai4bharat/Rasa · Paper: arXiv 2407.14056