| --- |
| license: apache-2.0 |
| datasets: |
| - simon3000/genshin-voice |
| language: |
| - zh |
| base_model: |
| - SparkAudio/Spark-TTS-0.5B |
| pipeline_tag: text-to-speech |
| tags: |
| - chinese |
| - spark-tts |
| - genshin |
| - float16 |
| --- |
| # Spark TTS finetuned in genshin charactors voices. |
| ## available charactors |
| * paimon, hutao, furina, kazuha, xiao, mona, ganyu, xiangling, shotgun, citlali.(each charactor in one full finetuned model) |
| # Usage |
| ### Models download |
| ``` |
| from huggingface_hub import hf_hub_download |
| |
| # Download furina model as example |
| repo="wesjos/spark-tts-genshin-charactors" |
| charactor='furina' ## |
| model_path = hf_hub_download( |
| repo_id=repo, |
| filename=charactor, |
| repo_type="model" |
| ) |
| |
| audio_tokenizer_path= hf_hub_download( |
| repo_id=repo, |
| filename="Spark-TTS-0.5B", |
| repo_type="model" |
| ) |
| |
| print(f'save model to {model_path}') |
| print(f'save audio_tokenizer to {audio_tokenizer_path}') |
| ``` |
| # Inference usage |
| * see code in ('https://github.com/nonwesjoe/genshin-sparktts') |
| * audio ouputs examples: (https://github.com/nonwesjoe/genshin-sparktts/tree/main/examples) |
| # Detail |
| * this model is trianed on float32 but saved as float16 for less VRAM and Storage usage. |
| # Acknowledgement |
| * [Spark-TTS](https://github.com/SparkAudio/Spark-TTS) |
| * [Genshin dataset](https://huggingface.co/datasets/simon3000/genshin-voice) |
| * [Unsloth](https://github.com/unslothai/unsloth) |
