README.md

---
license: cc-by-nc-4.0
language:
- id
base_model:
- SWivid/F5-TTS
pipeline_tag: text-to-speech
---

## Overview  
This indonesian finetune of [F5-TTS](https://github.com/SWivid/F5-TTS) is made to introduce indonesian speech capabilities on the model.

## Dataset  
Length: 43.35 hours \
Audio samples: 43999

Dataset sources: \
• [data-indsp-news-lvcsr](https://github.com/s-sakti/data_indsp_news_lvcsr)

## Results  
The model has some difficulties in accurately matching the zero shot voice and emotions. The model also hallucinates on long texts.

Reference text: "Tidak ada yang menakutiku, bahkan kematian sekalipun." \
Reference audio: [Zilong.ogg](https://huggingface.co/Eempostor/F5-TTS-IND-FINETUNE/resolve/main/Zilong.ogg?download=true) \
Input text: "Halo. Model faintun ini adalah sebuah percobaan. Masih terdapat beberapa kekurangan jadi tolong dimaklumkan." \
Generated audio: [Zilong_generated.ogg](https://huggingface.co/Eempostor/F5-TTS-IND-FINETUNE/resolve/main/Zilong_generated.wav?download=true)

## License  
The pre-trained models are licensed under the CC-BY-NC license due to the training data Emilia, which is an in-the-wild dataset. Sorry for any inconvenience this may cause.

---