Fine-tune coqui XTTS model for Taiwanese accent

[wkCircle]

Existing Status & Motivation

After some surveying in the previous ticket #8 , coqui xTTS model looks the most promising direction to go for Taiwanese accent downloading, since it supports multi-languages, it is open sourced that allows for commercial use, and it has good performance quality than Bark, and MS-TTS model.

Goal

trained a fine-tuned xTTS model on Taiwanese dataset and review the result of synthesized output audios.

References

• Coqui Discord forum
• Coqui Documentation for nearly everything including training, fine-tuning, inferencing, datasets etc.

[wkCircle]

On hold on 2024/4/14 and decide to firstly implement the framework towards the chaining of model A,B,C and for minimal MVP.