|
--- |
|
library_name: fastpitch |
|
task: text-to-speech |
|
tags: |
|
- fastpitch |
|
- waveglow |
|
- text-to-speech |
|
language: rw |
|
datasets: |
|
- mbazaNLP/kinyarwanda-tts-dataset |
|
widget: |
|
- text: "Muraho neza, murakaza neza mu Rwanda." |
|
example_title: "Muraho neza, murakaza neza mu Rwanda." |
|
--- |
|
|
|
**Model card - Kinyarwanda TTS model** |
|
|
|
**Model details** |
|
- Kinyarwanda Text to Speech model |
|
- Developed by [Digital Umuganda](digitalumuganda.com), [Arxia](https://www.arxia.com/home.html) and [Zevo Tech](https://zevo-tech.com/) |
|
- Model based from: Fastspeech and Waveglow |
|
- License: Mozilla 2.0 License |
|
- Feedback on the model: [email protected] |
|
|
|
**Metrics** |
|
- We use Mean Opinion Score (MOS) to evaluate the model with a maximum score being 5 |
|
|
|
|Test Corpus|MOS| |
|
|-----------|---| |
|
|Custom phrases|3| |
|
|
|
**Challenges** |
|
- The model does not always capture the Kinyarwanda tones |
|
|
|
**Recommendations** |
|
- Use a tonal dictionary to train future models |
|
- Add a numbers and symbols Dictionary |
|
- Create a code-switching dictionary containing foreign words used in Kinyarwanda |
