Run it yourself
Requirements
Section titled “Requirements”| Minimum | Tested on | |
|---|---|---|
| GPU | NVIDIA, 6 GB VRAM (CUDA) | RTX 3050 Laptop 6 GB |
| RAM | 8 GB | 16 GB |
| Disk | ~10 GB (environment + model weights) | |
| Software | Python 3.10, FFmpeg on PATH, Node.js 22+ | Windows 11 |
-
Clone the repository
Terminal window git clone https://github.com/Aayush-Dubey123/audiopipeline.gitcd audiopipeline -
Create the Python environment (one environment, inside
backend/)Terminal window python -m venv backend\.venvbackend\.venv\Scripts\Activate.ps1Terminal window python -m venv backend/.venvsource backend/.venv/bin/activate -
Install PyTorch with CUDA, then the project
Terminal window pip install torch==2.6.0+cu124 torchaudio==2.6.0+cu124 --extra-index-url https://download.pytorch.org/whl/cu124pip install -r requirements.txt -
Configure by copying
backend/.env.exampletobackend/.env. The defaults work; the voice-tuning settings are explained in Tuning by measurement. -
Start everything
Terminal window .\run.ps1This opens the backend on
http://127.0.0.1:8000(API docs at/docs) and the studio onhttp://localhost:3000. The first generation downloads the model weights (a few GB) from Hugging Face.
Run the tests
Section titled “Run the tests”backend\.venv\Scripts\python.exe -m pytest backend/tests -qThe suite uses a throwaway database and storage folder, so it leaves nothing behind.
Tune the voice
Section titled “Tune the voice”All three settings live in backend/.env:
CHATTERBOX_EXAGGERATION=0.5 # expressiveness; keep ≤ 0.6 to avoid babblingCHATTERBOX_CFG_WEIGHT=0.4 # lower follows the reference's pace and accent more closelyCHATTERBOX_TEMPERATURE=0.75 # higher = more variation between chunks