One-click Pinokio launcher for DocToSpeech with Sesame CSM-1B.
| Model | VRAM | Cloning | Languages | Presets |
|---|---|---|---|---|
| Sesame CSM-1B | ~4.5 GB (CUDA) / ~8.5 GB (CPU) | Yes | English | No |
Conversational speech model with Llama backbone. Requires CUDA 12.x and a HuggingFace account with gated model access.
- Python 3.11
- 8 GB RAM minimum, 16 GB+ recommended
- 4.5 GB GPU VRAM (CUDA 12.x required)
- Pinokio
- HuggingFace account (for gated model access)
Poppler is required for PDF text extraction on all platforms. If you plan to process PDF files, you must install Poppler separately — it is not included with pip dependencies.
| Platform | Install Command |
|---|---|
| Linux (Debian/Ubuntu) | sudo apt install poppler-utils |
| Linux (Fedora/RHEL) | sudo dnf install poppler-utils |
| Linux (Arch) | sudo pacman -S poppler |
| macOS | brew install poppler |
| Windows | Download from poppler-windows releases, extract, and add the Library/bin folder to your system PATH |
After installing, verify with: pdftotext -v
Without Poppler, PDF documents will fail to load in the Gradio web UI and CLI.
- Open Pinokio
- Click Discover
- Search for DocToSpeech (CSM)
- Click Download then Install
Or manually copy this folder to your Pinokio api/ directory.
Important: Before starting the app, run huggingface-cli login in your terminal to authenticate. CSM models require gated access from HuggingFace.
After installation, Pinokio will show a Start button. Click it to launch the Gradio web UI at http://localhost:7860.
| Launcher | Models | Venv |
|---|---|---|
| DocToSpeech (Coqui) | XTTS v2, Bark, VITS, YourTTS, StyleTTS 2 | Shared |
| DocToSpeech (Chatterbox) | Resemble Chatterbox | Separate |
| DocToSpeech (Tortoise) | Tortoise TTS | Separate |
| DocToSpeech (OpenVoice) | OpenVoice V2 | Separate |
| DocToSpeech (CSM) (this one) | Sesame CSM-1B | Separate |
For CLI usage, batch processing, Ollama sanitization, and more, see the main DocToSpeech repository.
Started on July 12th, 2026 by C0m3b4ck.
- Sesame CSM by Sesame AI -- Apache 2.0 License
APL 2.0 License.