readme: | # Fish Diffusion - HiFiSinger Demo 🎤 GitHub Repo: [fishaudio/fish-diffusion](https://github.com/fishaudio/fish-diffusion) To share a new model, please check out the [Share Your Model](https://huggingface.co./spaces/fishaudio/fish-diffusion/discussions/2) discussion. max_mixing_speakers: 3 models: - name: "M4Singer Pretrained (Many Speakers, Alto, Tenor, Soprano, Bass)" config: configs/exp_hifisinger.py checkpoint: checkpoints/checkpoint.ckpt readme: | This model is trained on the Opencpop and M4Singer dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It contains more than 20 speakers and is thus a good playground for timbre mixing. default_speaker: "opencpop" - name: "Tohoku Kiritan (Feminine)" config: configs/Kiritan.py checkpoint: checkpoints/Kiritan.ckpt readme: | This model is trained on the Tohoku Kiritan dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It has a cute, yet powerful voice. default_speaker: "kiritan" - name: "Tohoku Itako (Feminine)" config: configs/Itako.py checkpoint: checkpoints/Itako.ckpt readme: | This model is trained on the Tohoku Itako dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It has a bright and whispery voice. default_speaker: "itako" - name: "No.7 (Feminine)" config: configs/Seven.py checkpoint: checkpoints/Seven.ckpt readme: | This model is trained on the No.7 dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It has a strong and sharp voice. default_speaker: "seven" - name: "Yoko (Feminine)" config: configs/Yoko.py checkpoint: checkpoints/Yoko.ckpt readme: | This model is trained on the Sinsy-f00001 dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It has a powerful, tense, and relaxed voice. default_speaker: "yoko" - name: "JSUT (Feminine)" config: configs/JSUT.py checkpoint: checkpoints/JSUT.ckpt readme: | This model is trained on the JSUT-song dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It has a moist and transparent voice. default_speaker: "jsut" - name: "CSD (Feminine)" config: configs/CSD.py checkpoint: checkpoints/CSD.ckpt readme: | This model is trained on the Children's Song Dataset and released under the [CC-BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. It has a sweet and tender voice. default_speaker: "csd"