CSM Sample

csm.py

from huggingface_hub import hf_hub_download
from generator import load_csm_1b
import torchaudio

# need this to access to model. 
# you also need to get access to https://huggingface.co/meta-llama/Llama-3.2-1B
from huggingface_hub import login
login("your_huggingface_token")

model_path = hf_hub_download(repo_id="sesame/csm-1b", filename="ckpt.pt")
generator = load_csm_1b(model_path, "cuda")
audio = generator.generate(
    text="Hello from Sesame.",
    speaker=0,
    context=[],
    max_audio_length_ms=10_000,
)

torchaudio.save("audio.wav", audio.unsqueeze(0).cpu(), generator.sample_rate)

https://github.com/SesameAILabs/csm

git clone [email protected]:SesameAILabs/csm.git
cd csm
python3.10 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

make sure you have ffmpeg installed as well

sudo apt update
sudo apt install ffmpeg