VASTAI-STUDIO-AUDIO/install-petitsmodelesvastai.md

cat << 'EOF' > /workspace/REINSTALL_CLEAN.md

📓 IA WORKSPACE : BACKUP COMPLET

🎤 1. SYNTHÈSE VOCALE (PIPER TTS)

Installation Système

apt-get update && apt-get install -y espeak-ng-data libespeak-ng1 ffmpeg wget

Installation Python (venv)

source /venv/main/bin/activate pip install piper-tts flask uvicorn requests

Téléchargement Modèles

cd /workspace wget -O voice.onnx https://huggingface.co/rhasspy/piper-voices/resolve/main/fr/fr_FR/siwis/low/fr_FR-siwis-low.onnx wget -O voice.onnx.json https://huggingface.co/rhasspy/piper-voices/resolve/main/fr/fr_FR/siwis/low/fr_FR-siwis-low.onnx.json

🎨 2. GÉNÉRATION D'IMAGES (COMFYUI)

cd /workspace git clone https://github.com/comfyanonymous/ComfyUI.git cd ComfyUI pip install -r requirements.txt

Modèle SDXL

cd /workspace/ComfyUI/models/checkpoints/ wget https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0.safetensors

🔗 3. URL OPEN WEBUI

• TTS: http://localhost:8000/v1 (Modèle: voice.onnx)
• IMAGE: http://localhost:8188 EOF