Cog implementation of MusicGen

MusicGen is a simple and controllable model for music generation. It is a single stage auto-regressive Transformer model trained over a 32kHz EnCodec tokenizer with 4 codebooks sampled at 50 Hz. Unlike existing methods like MusicLM, MusicGen doesn't require a self-supervised semantic representation, and it generates all 4 codebooks in one pass. By introducing a small delay between the codebooks, the authors show they can predict them in parallel, thus having only 50 auto-regressive steps per second of audio. They used 20K hours of licensed music to train MusicGen. Specifically, they relied on an internal dataset of 10K high-quality music tracks, and on the ShutterStock and Pond5 music data.

For more information about this model, see here.

You can demo this model or learn how to use it with Replicate's API here.

Run with Cog

Cog is an open-source tool that packages machine learning models in a standard, production-ready container. You can deploy your packaged model to your own infrastructure, or to Replicate, where users can interact with it via web interface or API.

Prerequisites

Cog. Follow these instructions to install Cog, or just run:

sudo curl -o /usr/local/bin/cog -L "https://github.com/replicate/cog/releases/latest/download/cog_$(uname -s)_$(uname -m)"
sudo chmod +x /usr/local/bin/cog

Note, to use Cog, you'll also need an installation of Docker.

• GPU machine. You'll need a Linux machine with an NVIDIA GPU attached and the NVIDIA Container Toolkit installed. If you don't already have access to a machine with a GPU, check out our guide to getting a GPU machine.

Step 1. Clone this repository

git clone https://github.com/replicate/cog-musicgen-melody

Step 2. Run the model

To run the model, you need a local copy of the model's Docker image. You can satisfy this requirement by specifying the image ID in your call to predict like:

cog predict r8.im/joehoover/musicgen-melody@sha256:1a53415e6c4549e3022a0af82f4bd22b9ae2e747a8193af91b0bdffe63f93dfd -i description=tense staccato strings. plucked strings. dissonant. scary movie. -i duration=8

For more information, see the Cog section here

Alternatively, you can build the image yourself, either by running cog build or by letting cog predict trigger the build process implicitly. For example, the following will trigger the build process and then execute prediction:

cog predict -i description="tense staccato strings. plucked strings. dissonant. scary movie." -i duration=8

Note, the first time you run cog predict, model weights and other requisite assets will be downloaded if they're not available locally. This download only needs to be executed once.

Run on replicate

Step 1. Ensure that all assets are available locally

If you haven't already, you should ensure that your model runs locally with cog predict. This will guarantee that all assets are accessible. E.g., run:

cog predict -i description=tense staccato strings. plucked strings. dissonant. scary movie. -i duration=8

Step 2. Create a model on Replicate.

Go to replicate.com/create to create a Replicate model. If you want to keep the model private, make sure to specify "private".

Step 3. Configure the model's hardware

Replicate supports running models on variety of CPU and GPU configurations. For the best performance, you'll want to run this model on an A100 instance.

Click on the "Settings" tab on your model page, scroll down to "GPU hardware", and select "A100". Then click "Save".

Step 4: Push the model to Replicate

Log in to Replicate:

cog login

Push the contents of your current directory to Replicate, using the model name you specified in step 1:

cog push r8.im/username/modelname

Learn more about pushing models to Replicate.

Fine-tune MusicGen

Support for fine-tuning MusicGen is in development. Currently, minimal support has been implemented via an adaptation of @chavez's music_gen trainer.

Assuming you have a local environment configured (i.e. you've completed the steps specified under Run with Cog), you can run training with a command like:

cog train -i dataset_path=@<path-to-your-data> <additional hyperparameters>

Data preparation for training

Cog requires input data to be a file; however, our training script expects a directory. Accordingly, in production, training data should be provided as a tarball of a directory of properly formatted training data. However, you can bypass this requirement by naming your training data directory ./train_data. If such a directory exists, the training script will attempt to load data from that directory (see lines 140-147 in train.py).

Currently, training only supports music generation with text prompts.

To train the model on your own data, follow these steps:

1. Convert your audio files to .wav segments of no more than 30 seconds'
2. Every audio file in your training directory must have a correspondint .txt file with the same filename. These text files should contain the text prompt that you want to associat with the corresponding audio file. For example, if you have audio_1.wav, you must also have audio_1.txt and that text file should contain the prompt for audio_1.wav.
3. These files should be placed in a single directory.
4. If that directory is called ./train_data, then you can simply run the training script like:

cog train -i dataset_path=@./train_data/ <additional hyperparameters>

5. Alternatively, if train_data does not exist, you can tarball your data directory and pass the path to the tarball to cog train .... The train script will then untar your data and attempt to load it.

Example

Run this to train on a single clip:

mkdir ./train_data/
wget -P ./train_data/ https://github.com/facebookresearch/audiocraft/raw/main/assets/bach.mp3
echo bach > ./train_data/bach.txt
tar -cvzf train_data.tar.gz train_data/
cog train -i dataset_path=@./data.tar.gz -i epochs=10

Then, you can load your model like model.lm.load_state_dict(torch.load('model_outdir/lm_final.pt')) and generate like:

model.set_generation_params(
    duration=8,
    top_k=250,
    top_p=0,
    temperature=1,
    cfg_coef=3,
)
wav = model.generate(descriptions=[''], progress=True)

Licenses

• All code in this repository is licensed under the Apache License 2.0 license.
• The code in the Audiocraft repository is released under the MIT license as found in the LICENSE file.
• The weights in the Audiocraft repository are released under the CC-BY-NC 4.0 license as found in the LICENSE_weights file.