| --- |
| license: gpl-2.0 |
| language: |
| - hy |
| tags: |
| - armenian |
| - text-to-speech |
| - tts |
| - speech |
| - omnivoice |
| - pytorch |
| pipeline_tag: text-to-speech |
| --- |
| |
| # AVoice-TTS |
|
|
| Open source Armenian text-to-speech checkpoint for AVoice. |
|
|
| Runtime code: |
|
|
| `https://github.com/Hakob-Harutyunyan/AVoice` |
|
|
| ## Usage |
|
|
| Try it in Kaggle! |
| https://www.kaggle.com/code/hakdevelopment/avoice-testing |
|
|
| ```bash |
| git clone https://github.com/Hakob-Harutyunyan/AVoice.git |
| cd Avoice |
| pip install -U pip |
| pip install -e . |
| |
| omnivoice-infer \ |
| --model Hak5/AVoice-TTS \ |
| --text "Բարեւ աշխարհ։" \ |
| --language hy \ |
| --duration 4 \ |
| --output sample.wav |
| ``` |
|
|
| ## Docker |
|
|
| ```bash |
| docker build -t avoice-api https://github.com/Hakob-Harutyunyan/AVoice.git |
| |
| docker run --gpus all --rm -p 8000:8000 \ |
| -e OMNIVOICE_MODEL=Hak5/AVoice-TTS \ |
| -e OMNIVOICE_DEVICE=cuda \ |
| avoice-api |
| ``` |
|
|
|
|
