Fish Speech Howto

Install dependencies: Install required packages by running: pip3 install torch torchvision torchaudio

Create virtual environment: Create a Python 3.10 virtual environment using conda: conda create -n fish-speech python=3.10

Activate environment: Activate the virtual environment: conda activate fish-speech

Install Fish Speech: Install Fish Speech by running: pip3 install -e .

Download models: Download required models from Hugging Face: huggingface-cli download fishaudio/fish-speech-1.2-sft --local-dir checkpoints/fish-speech-1.2-sft

Run inference: Generate speech by running: python tools/llama/generate.py --text "Your text here" --checkpoint-path "checkpoints/fish-speech-1.2-sft"

Decode audio: Decode the generated tokens to audio using VQGAN: python tools/vqgan/inference.py -i "codes_0.npy" --checkpoint-path "checkpoints/fish-speech-1.2-sft/firefly-gan-vq-fsq-4x1024-42hz-generator.pth"

Start web UI (optional): Launch the web interface by running: python -m tools.webui --llama-checkpoint-path "checkpoints/fish-speech-1.2-sft" --decoder-checkpoint-path "checkpoints/fish-speech-1.2-sft/firefly-gan-vq-fsq-4x1024-42hz-generator.pth"