Supernova TeraiTTS
Nepali text-to-speech for Python.
Supernova TeraiTTS is a Python library designed to make Nepali speech generation accessible through a simple API. it uses Supernova11c founders original voice for speaking
Current implementation
The current implementation uses the Nepali XTTS v2 fine-tune from Oshara/xtts-v2-nepali. Supernova TeraiTTS provides a Python interface around this upstream model; the library and underlying model are separate components.
Installation
pip install supernova-teraitts
Usage
from supernova_terai import tts
audio = tts.generate(
"नमस्ते, तपाईंलाई कस्तो छ?",
speaker_wav="my_voice.wav",
output_path="hello.wav",
)
audio.play()
Model details
- Task: text-to-speech
- Primary language: Nepali
- Output sample rate: 24 kHz
- Upstream model: Oshara/xtts-v2-nepali
- Base architecture: Coqui XTTS v2
Model weights and licensing
The current library downloads the upstream model at runtime. This repository does not claim the upstream weights as independently trained Supernova weights. Any separate Supernova checkpoint must be identified and documented explicitly. Review the upstream license before redistributing model weights. Specify a separate license for the library code.
Limitations
Speech quality may vary with text length, pronunciation, reference-audio quality, and hardware. Only use voice recordings you have permission to use.
Project status
Early release. Testing and evaluation are ongoing.
Maintainer
Supernova11c
Important notice
many people may thought that supernova is fine tuning while if we speak about exact truth that is to understand the technology we have done fine tuning our next model would include scratch made as following as tokenizer