multimodalart's picture
multimodalart HF Staff
Upload folder using huggingface_hub
fdcef0f verified
|
Raw
History Blame Contribute Delete
1.4 kB
---
title: Irodori-TTS-v4.1-Small INT8 Quantized Demo
emoji: 🗣️
colorFrom: pink
colorTo: gray
sdk: gradio
sdk_version: 6.22.0
python_version: '3.12'
app_file: app.py
short_description: Japanese TTS with INT8-quantized Irodori-TTS-v4.1-Small
startup_duration_timeout: 30m
license: mit
models:
- Aratako/Irodori-TTS-v4.1-Small-Quantized
- Aratako/Irodori-TTS-v4.1-Small
- Aratako/Semantic-DACVAE-Japanese-32dim
---
# Irodori-TTS-v4.1-Small (INT8 Quantized) Demo
This Space demonstrates **Irodori-TTS-v4.1-Small** loaded from the **INT8 weight-only quantized** variant (`Aratako/Irodori-TTS-v4.1-Small-Quantized/int8-weight-only`), which reduces the model footprint while preserving quality.
## Features
- **Text-to-Speech**: Generate Japanese speech from text.
- **Voice Cloning**: Upload reference audio clips to clone a speaker's voice.
- **Style Control**: Use the caption field for emotion, tone, and speaking style prompts.
- **Emoji Palette**: Insert control emojis to influence speech delivery.
## Model
- [Quantized Model](https://huggingface.co/Aratako/Irodori-TTS-v4.1-Small-Quantized)
- [Base Model](https://huggingface.co/Aratako/Irodori-TTS-v4.1-Small)
- [GitHub](https://github.com/Aratako/Irodori-TTS)
- [Codec](https://huggingface.co/Aratako/Semantic-DACVAE-Japanese-32dim)
## License
MIT — subject to ethical restrictions regarding impersonation and misinformation.