Module `audiocraft.models`

Models for EnCodec, AudioGen, MusicGen, as well as the generic LMModel.

Expand source code

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
"""
Models for EnCodec, AudioGen, MusicGen, as well as the generic LMModel.
"""
# flake8: noqa
from . import builders, loaders
from .encodec import (
    CompressionModel, EncodecModel, DAC,
    HFEncodecModel, HFEncodecCompressionModel)
from .audiogen import AudioGen
from .lm import LMModel
from .lm_magnet import MagnetLMModel
from .multibanddiffusion import MultiBandDiffusion
from .musicgen import MusicGen
from .magnet import MAGNeT
from .unet import DiffusionUnet

Sub-modules

audiocraft.models.audiogen: Main model for using AudioGen. This will combine all the required components and provide easy access to the generation API.
audiocraft.models.builders: All the functions to build the relevant models and modules from the Hydra config.
audiocraft.models.encodec: Compression models or wrapper around existing models. Also defines the main interface that a model must follow to be usable as an audio tokenizer.
audiocraft.models.genmodel: Base implementation for audio generative models. This base implementation combines all the required components to run inference with pretrained audio …
audiocraft.models.lm
audiocraft.models.lm_magnet
audiocraft.models.loaders: Utility functions to load from the checkpoints. Each checkpoint is a torch.saved dict with the following keys: - 'xp.cfg': the hydra config as dumped …
audiocraft.models.magnet: Main model for using MAGNeT. This will combine all the required components and provide easy access to the generation API.
audiocraft.models.multibanddiffusion: Multi Band Diffusion models as described in "From Discrete Tokens to High-Fidelity Audio Using Multi-Band Diffusion" (paper link).
audiocraft.models.musicgen: Main model for using MusicGen. This will combine all the required components and provide easy access to the generation API.
audiocraft.models.unet: Pytorch Unet Module used for diffusion.