Audiocraft
audiogen
Edit model card

AudioGen - Medium - 1.5B

AudioGen is an autoregressive transformer LM that synthesizes general audio conditioned on text (Text-to-Audio). Internally, AudioGen operates over discrete representations learnt from the raw waveform, using an EnCodec tokenizer.

AudioGen was presented at AudioGen: Textually Guided Audio Generation by Felix Kreuk, Gabriel Synnaeve, Adam Polyak, Uriel Singer, Alexandre DΓ©fossez, Jade Copet, Devi Parikh, Yaniv Taigman, Yossi Adi.

AudioGen 1.5B is a variant of the original AudioGen model that follows MusicGen architecture. More specifically, it is trained over a 16kHz EnCodec tokenizer with 4 codebooks sampled at 50 Hz with a delay pattern between the codebooks. Having only 50 auto-regressive steps per second of audio, this AudioGen model allows faster generation while reaching similar performances to the original AudioGen model introduced in the paper.

Audiocraft Usage

You can run AudioGen locally through the original Audiocraft library:

  1. First install the audiocraft library
pip install git+https://github.com/facebookresearch/audiocraft.git
  1. Make sure to have ffmpeg installed:
apt get install ffmpeg
  1. Run the following Python code:
import torchaudio
from audiocraft.models import AudioGen
from audiocraft.data.audio import audio_write

model = AudioGen.get_pretrained('facebook/audiogen-medium')
model.set_generation_params(duration=5)  # generate 5 seconds.
descriptions = ['dog barking', 'sirenes of an emergency vehicule', 'footsteps in a corridor']
wav = model.generate(descriptions)  # generates 3 samples.

for idx, one_wav in enumerate(wav):
    # Will save under {idx}.wav, with loudness normalization at -14 db LUFS.
    audio_write(f'{idx}', one_wav.cpu(), model.sample_rate, strategy="loudness", loudness_compressor=True)

Model details

See AudioGen's model card.

Downloads last month
5,501
Inference API
Unable to determine this model’s pipeline type. Check the docs .

Spaces using facebook/audiogen-medium 100