Music Source Separation in the Waveform Domain

We provide an implementation of Demucs and Conv-Tasnet for music source separation on the MusDB dataset. They can separate drums, bass and vocals from the rest with state-of-the-art results, surpassing previous waveform or spectrogram based methods. The architecture and results obtained are detailed in our paper Music Source Separation in the waveform domain.

Demucs is based on U-Net convolutional architecture inspired by Wave-U-Net and SING, with GLUs, a BiLSTM between the encoder and decoder, specific initialization of weights and transposed convolutions in the decoder.

Conv-Tasnet is a separation model developed for speech which predicts a mask on a learnt over-complete linear representation using a purely convolutional model with stride of 1 and dilated convolutional blocks. We reused the code from the kaituoxu/Conv-TasNet repository and added support for multiple audio channels.

When trained only on MusDB, Conv-Tasnet achieves higher SDR than Demucs (5.7 vs 5.6). However, the audio it generates has significant artifacts as measured by human evaluations (MOS is 3.2 for Demucs, 2.9 for Conv-Tasnet). When trained with extra training data, Demucs and Conv-Tasnet obtain the same SDR. See our paper Section 6 for more details or listen to our audio samples .

Comparison with other models

An audio comparison of Demucs and Conv-Tasnet with other state-of-the-art methods such as Wave-U-Net, OpenUnmix or MMDenseLSTM is available on the audio comparison page. We provide hereafter a summary of the different metrics presented in the paper. You can also compare Spleeter, Open-Unmix, Demucs and Conv-Tasnet on one of my favorite songs on our soundcloud playlist.

Comparison of accuracy

Overall SDR is the mean of the SDR for each of the 4 sources, MOS Quality is a rating from 1 to 5 of the naturalness and absence of artifacts given by human listeners (5 = no artifacts), MOS Contamination is a rating from 1 to 5 with 5 being zero contamination by other sources. We refer the reader to our paper, Section 5 and 6, for more details.

Model	Domain	Extra data?	Overall SDR	MOS Quality	MOS Contamination
Open-Unmix	spectrogram	no	5.3	3.0	3.3
Wave-U-Net	waveform	no	3.2	-	-
Demucs (this)	waveform	no	5.6	3.2	3.3
Conv-Tasnet (this)	waveform	no	5.7	2.9	3.4
Demucs (this)	waveform	150 songs	6.3	-	-
Conv-Tasnet (this)	waveform	150 songs	6.3	-	-
MMDenseLSTM	spectrogram	804 songs	6.0	-	-
Spleeter	spectrogram	undisclosed	5.9	-	-

Requirements

If you have anaconda installed, you can run from the root of this repository:

conda env update -f environment-cpu.yml # if you don't have GPUs
conda env update -f environment-cuda.yml # if you have GPUs
conda activate demucs

This will create a demucs environment with all the dependencies installed.

Using Windows

If you are using Windows, replace python3 by python.exe in all the commands provided hereafter :) Parts of the code are untested on Windows (in particular, training a new model). If you don't have much experience with Anaconda, python or the shell, here are more detailed instructions. Note that Demucs is not supported on 32bits systems (as Pytorch is not available there).

First install Anaconda with Python 3.7, which you can find here.
Start the Anaconda prompt.
Type in the following commands:

cd %HOMEPATH%
conda install git
git clone https://github.com/facebookresearch/demucs
cd demucs
conda env update -f environment-cpu.yml
conda activate demucs
python.exe -m demucs.separate --dl "PATH_TO_AUDIO_FILE_1" ["PATH_TO_AUDIO_FILE_2" ...]

The " around the filename are required if the path contains spaces. The separated files will be under C:\Users\YOUR_USERNAME\demucs\separated\demucs\. The next time you want to use Demucs, start again the Anaconda prompt and simply run

cd %HOMEPATH%
cd demucs
python.exe -m demucs.separate --dl "PATH_TO_AUDIO_FILE_1" ...

If you have an error saying that mkl_intel_thread.dll cannot be found, you can try to first run set CONDA_DLL_SEARCH_MODIFICATION_ENABLE=1 and hopefully it will work 🙏.

Separating tracks

In order to try Demucs or Conv-Tasnet on your tracks, simply run from the root of this repository

python3 -m demucs.separate --dl -n demucs PATH_TO_AUDIO_FILE_1 [PATH_TO_AUDIO_FILE_2 ...] # for Demucs
python3 -m demucs.separate --dl -n tasnet PATH_TO_AUDIO_FILE_1 ... # for Conv-Tasnet
# Demucs with randomized equivariant stabilization (10x slower, suitable for GPU, 0.2 extra SDR)
python3 -m demucs.separate --dl -n demucs --shifts=10 PATH_TO_AUDIO_FILE_1

If you have a GPU, but you run out of memory, please add -d cpu to the command line. See the section hereafter for more details on the memory requirements for GPU acceleration.

The --dl flag will automatically download a pretrained model into ./models. There will be one folder per audio file, reusing the name of the track without the extension. Each folder will contain four stereo wav files sampled at 44.1 kHz: drums.wav, bass.wav, other.wav, vocals.wav. Those folders will be placed in ./separated/MODEL_NAME.

Any stereo audio file supported by ffmpeg will work. It will be resampled to 44.1 kHz on the fly if necessary. If multiple streams (i.e. a stems file) are present in the audio file, the first one will be used.

Other pre-trained models can be selected with the -n flag and downloaded with the --dl flag. The models will be stored in the models folder. The list of pre-trained models is:

demucs: Demucs trained on MusDB,
demucs_extra: Demucs trained with extra training data,
tasnet: Conv-Tasnet trained on MusDB,
tasnet_extra: Conv-Tasnet trained with extra training data.

The --shifts=SHIFTS performs multiple predictions with random shifts (a.k.a randomized equivariant stabilization) of the input and average them. This makes prediction SHIFTS times slower but improves the accuracy of Demucs by 0.2 points of SDR. It has limited impact on Conv-Tasnet as the model is by nature almost time equivariant. The value of 10 was used on the original paper, although 5 yields mostly the same gain. It is deactivated by default.

Memory requirements for GPU acceleration

If you want to use GPU acceleration, you will need at least 8GB of RAM on your GPU for demucs and 4GB for tasnet. Sorry, the code for demucs is not super optimized for memory! I'll publish soon a lighter version of the model that should run with less RAM. IF you do not have enough memory on your GPU, simply add -d cpu to the command line to use the CPU. With Demucs, processing time should be roughly equal to the duration of the track.

Examining the results from the paper experiments

The metrics for our experiments are stored in the results folder. In particular museval json evaluations are stored in results/evals/EXPERIMENT NAME/results. You can aggregate and display the results using

python3 valid_table.py -p # show valid loss, aggregated with multiple random seeds
python3 result_table.py -p # show SDR on test set, aggregated with multiple random seeds
python3 result_table.py -p SIR # also SAR, ISR, show other metrics

The std column shows the standard deviation divided by the square root of the number of runs.

Training Demucs and evaluating on the MusDB dataset

If you want to train Demucs from scratch, you will need a copy of the MusDB dataset. It can be obtained on the MusDB website. To start training on a single GPU or CPU, use:

python3 -m demucs -b 4  --musdb MUSDB_PATH # Demucs
python3 -m demucs -b 4  --musdb MUSDB_PATH --tasnet --samples=80000 --split_valid # Conv-Tasnet

The -b 4 flag will set the batch size to 4. The default is 4 and will crash on a single GPU. Demucs was trained on 8 V100 with 32GB of RAM. The default parameters (batch size, number of channels etc) might not be suitable for 16GB GPUs. To train on all available GPUs, use:

python3 run.py --musdb MUSDB_PATH [EXTRA_FLAGS]

This will launch one process per GPU and report the output of the first one. When interrupting such a run, it is possible some of the children processes are not killed properly, be mindful of that. If you want to use only some of the available GPUs, export the CUDA_VISIBLE_DEVICES variable to select those.

To see all the possible options, use python3 -m demucs --help.

About checkpointing

Demucs will automatically generate an experiment name from the command line flags you provided. It will checkpoint after every epoch. If a checkpoint already exist for the combination of flags you provided, it will be automatically used. In order to ignore/delete a previous checkpoint, run with the -R flag. The optimizer state, the latest model and the best model on valid are stored. At the end of each epoch, the checkpoint will erase the one from the previous epoch. By default, checkpoints are stored in the ./checkpoints folder. This can be changed using the --checkpoints CHECKPOINT_FOLDER flag.

Not all options will impact the name of the experiment. For instance --workers is not shown in the name, therefore, changing this parameter will not impact the checkpoint file used. Refer to parser.py for more details.

Test set evaluations

Test set evaluations computed with museval will be stored under evals/EXPERIMENT NAME/results. The experiment name is the first thing printed when running python3 run.py or python3 -m demucs. If you used the flag --save, there will also be a folder evals/EXPERIMENT NAME/wavs containing all the extracted waveforms.

Running on a cluster

If you have a cluster available with Slurm, you can set the run_slurm.py as the target of a slurm job, using as many nodes as you want and a single task per node. run_slurm.py will create one process per GPU and run in a distributed manner. Multinode training is supported.

Extracting Raw audio for faster loading

We observed that loading from compressed mp4 audio lead to unreliable speed, sometimes reducing by a factor of 2 the number of iterations per second. It is possible to extract all data to raw PCM f32e format. If you wish to store the raw data under RAW_PATH, run the following command first:

python3 -m demucs.raw [--workers=10] MUSDB_PATH RAW_PATH

You can then train using the --raw RAW_PATH flag, for instance:

python3 run.py --raw RAW_PATH --musdb MUSDB_PATH

You still need to provide the path to the MusDB dataset as we always load the test set from the original MusDB.

Results reproduction

To reproduce the performance of the main Demucs model in our paper:

# Extract raw waveforms. This is optional
python3 -m demucs.data MUSDB_PATH RAW_PATH
export DEMUCS_RAW=RAW_PATH
# Train models with default parameters and multiple seeds
python3 run.py --seed 42 # for Demucs
python3 run.py --seed 42 --tasnet --X=10 --samples=80000 --epochs=180 --split_valid # for Conv-Tasnet
# Repeat for --seed = 43, 44, 45 and 46

You can visualize the results aggregated on multiple seeds using

python3 valid_table.py # compare validation losses
python3 result_table.py # compare test SDR
python3 result_table.py SIR # compare test SIR, also available ISR, and SAR

You can look at our exploration file dora.py to see the exact flags for all experiments (grid search and ablation study). If you have a Slurm cluster, you can also try adapting it to run on your own.

Environment variables

If you do not want to always specify the path to MUSDB, you can export the following variables:

export DEMUCS_MUSDB=PATH TO MUSDB
# Optionally, if you extracted raw pcm data
# export DEMUCS_RAW=PATH TO RAW PCM

How to cite

@techreport{music_separation_waveform,
  title = {{Music Source Separation in the Waveform Domain}},
  author = {D{\'e}fossez, Alexandre and Usunier, Nicolas and Bottou, L{\'e}on and Bach, Francis},
  year = {2019},
  number = {02379796v1},
  institution = {HAL},
}

License

Demucs is released under Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, as found in the LICENSE file.

The file demucs/tasnet.py is adapted from the kaituoxu/Conv-TasNet repository. It was originally released under the MIT License updated to support multiple audio channels.

molo32/demucs