forked from akhaliq

CODE_OF_CONDUCT.md

@@ -1,76 +0,0 @@
-# Code of Conduct
-
-## Our Pledge
-
-In the interest of fostering an open and welcoming environment, we as
-contributors and maintainers pledge to make participation in our project and
-our community a harassment-free experience for everyone, regardless of age, body
-size, disability, ethnicity, sex characteristics, gender identity and expression,
-level of experience, education, socio-economic status, nationality, personal
-appearance, race, religion, or sexual identity and orientation.
-
-## Our Standards
-
-Examples of behavior that contributes to creating a positive environment
-include:
-
-* Using welcoming and inclusive language
-* Being respectful of differing viewpoints and experiences
-* Gracefully accepting constructive criticism
-* Focusing on what is best for the community
-* Showing empathy towards other community members
-
-Examples of unacceptable behavior by participants include:
-
-* The use of sexualized language or imagery and unwelcome sexual attention or
-  advances
-* Trolling, insulting/derogatory comments, and personal or political attacks
-* Public or private harassment
-* Publishing others' private information, such as a physical or electronic
-  address, without explicit permission
-* Other conduct which could reasonably be considered inappropriate in a
-  professional setting
-
-## Our Responsibilities
-
-Project maintainers are responsible for clarifying the standards of acceptable
-behavior and are expected to take appropriate and fair corrective action in
-response to any instances of unacceptable behavior.
-
-Project maintainers have the right and responsibility to remove, edit, or
-reject comments, commits, code, wiki edits, issues, and other contributions
-that are not aligned to this Code of Conduct, or to ban temporarily or
-permanently any contributor for other behaviors that they deem inappropriate,
-threatening, offensive, or harmful.
-
-## Scope
-
-This Code of Conduct applies within all project spaces, and it also applies when
-an individual is representing the project or its community in public spaces.
-Examples of representing a project or community include using an official
-project e-mail address, posting via an official social media account, or acting
-as an appointed representative at an online or offline event. Representation of
-a project may be further defined and clarified by project maintainers.
-
-## Enforcement
-
-Instances of abusive, harassing, or otherwise unacceptable behavior may be
-reported by contacting the project team at <[email protected]>. All
-complaints will be reviewed and investigated and will result in a response that
-is deemed necessary and appropriate to the circumstances. The project team is
-obligated to maintain confidentiality with regard to the reporter of an incident.
-Further details of specific enforcement policies may be posted separately.
-
-Project maintainers who do not follow or enforce the Code of Conduct in good
-faith may face temporary or permanent repercussions as determined by other
-members of the project's leadership.
-
-## Attribution
-
-This Code of Conduct is adapted from the [Contributor Covenant][homepage], version 1.4,
-available at https://www.contributor-covenant.org/version/1/4/code-of-conduct.html
-
-[homepage]: https://www.contributor-covenant.org
-
-For answers to common questions about this code of conduct, see
-https://www.contributor-covenant.org/faq

CONTRIBUTING.md

@@ -1,23 +0,0 @@
-# Contributing to Demucs
-
-## Pull Requests
-
-In order to accept your pull request, we need you to submit a CLA. You only need
-to do this once to work on any of Facebook's open source projects.
-
-Complete your CLA here: <https://code.facebook.com/cla>
-
-Demucs is the implementation of a research paper.
-Therefore, we do not plan on accepting many pull requests for new features.
-We certainly welcome them for bug fixes.
-
-
-## Issues
-
-We use GitHub issues to track public bugs. Please ensure your description is
-clear and has sufficient instructions to be able to reproduce the issue.
-
-
-## License
-By contributing to this repository, you agree that your contributions will be licensed
-under the LICENSE file in the root directory of this source tree.

Demucs.ipynb

@@ -1,115 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "markdown",
-   "metadata": {
-    "colab_type": "text",
-    "id": "Be9yoh-ILfRr"
-   },
-   "source": [
-    "# [*Colab code for Demucs*](https://github.com/facebookresearch/demucs/)\n",
-    "\n",
-    "Original version by marlluslustosa **https://github.com/marlluslustosa/demucs/blob/master/Demucs.ipynb**\n",
-    "\n",
-    "However, now things are much simpler with Demucs v2, so this might not be so useful. There is now a Colab version:\n",
-    "https://colab.research.google.com/drive/1jCegIzLIuqqcM85uVs3WCeAJiSoYq3oh?usp=sharing"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "colab": {
-     "base_uri": "https://localhost:8080/",
-     "height": 139
-    },
-    "colab_type": "code",
-    "executionInfo": {
-     "elapsed": 12277,
-     "status": "ok",
-     "timestamp": 1583778134659,
-     "user": {
-      "displayName": "Marllus Lustosa",
-      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14GgLl2RbW64ZyWz3Y8IBku0zhHCMnt7fz7fEl0LTdA=s64",
-      "userId": "14811735256675200480"
-     },
-     "user_tz": 180
-    },
-    "id": "kOjIPLlzhPfn",
-    "outputId": "c75f17ec-b576-4105-bc5b-c2ac9c1018a3"
-   },
-   "outputs": [],
-   "source": [
-    "!pip install demucs"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {
-    "colab_type": "text",
-    "id": "Y1BdlzOQi3y7"
-   },
-   "source": [
-    "\n",
-    "\n",
-    "---\n",
-    "\n",
-    "\n",
-    "# **Here begins the code for separating the audio source (model pretrained)**\n",
-    "###**- Upload your song to demucs/ folder and edit YOUR-SONG-PATH.mp3**\n",
-    "\n",
-    "\n",
-    "---\n",
-    "\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "colab": {},
-    "colab_type": "code",
-    "id": "5lYOzKKCKAbJ"
-   },
-   "outputs": [],
-   "source": [
-    "!python3 -m demucs.separate test.mp3"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  }
- ],
- "metadata": {
-  "accelerator": "GPU",
-  "colab": {
-   "authorship_tag": "ABX9TyM9xpVr1M86NRcjtQ7g9tCx",
-   "collapsed_sections": [],
-   "name": "Demucs.ipynb",
-   "provenance": []
-  },
-  "kernelspec": {
-   "display_name": "Python 3",
-   "language": "python",
-   "name": "python3"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.8.3"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 1
-}

LICENSE

@@ -1,21 +0,0 @@
-MIT License
-
-Copyright (c) Facebook, Inc. and its affiliates.
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.

MANIFEST.in

@@ -1,6 +0,0 @@
-include *.md
-include LICENSE
-include setup.cfg
-incude demucs.png
-include requirements.txt
-recursive-include docs *.md

Makefile

@@ -1,19 +0,0 @@
-default: tests
-
-all: linter tests docs dist
-
-linter:
-	flake8 demucs
-
-tests:
-	python3 -m demucs.separate -n demucs_unittest test.mp3
-	python3 -m demucs.separate -n demucs_unittest --mp3 test.mp3
-
-dist:
-	python3 setup.py sdist
-
-clean:
-	rm -r dist build *.egg-info
-
-
-.PHONY: linter tests dist

README.md

@@ -1,379 +1,37 @@
-# Music Source Separation in the Waveform Domain
+---
+title: Demucs
+emoji: ⚡
+colorFrom: pink
+colorTo: indigo
+sdk: gradio
+app_file: app.py
+pinned: false
+---
 
-![tests badge](https://github.com/facebookresearch/demucs/workflows/tests/badge.svg)
-![linter badge](https://github.com/facebookresearch/demucs/workflows/linter/badge.svg)
+# Configuration
 
-**Branch was rename to main**: Run `git pull && git checkout main` to switch to the new branch.
+`title`: _string_  
+Display title for the Space
 
-**Demucs was just updated!**: much better SDR, smaller models, more data augmentation and PyPI support.
+`emoji`: _string_  
+Space emoji (emoji-only character allowed)
 
-**For the initial version of Demucs:** [Go this commit][original_demucs].
-If you are experiencing issues and want the old Demucs back, please fill an issue, and then you can get back to the v1 with
-`git checkout v1`.
+`colorFrom`: _string_  
+Color for Thumbnail gradient (red, yellow, green, blue, indigo, purple, pink, gray)
 
-We provide an implementation of Demucs and Conv-Tasnet for music source separation on the [MusDB][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_arxiv].
+`colorTo`: _string_  
+Color for Thumbnail gradient (red, yellow, green, blue, indigo, purple, pink, gray)
 
-Demucs is based on U-Net convolutional architecture inspired by [Wave-U-Net][waveunet] and
-[SING][sing], with GLUs, a BiLSTM between the encoder and decoder, specific initialization of weights
-and transposed convolutions in the decoder.
+`sdk`: _string_  
+Can be either `gradio` or `streamlit`
 
-[Conv-Tasnet](https://arxiv.org/abs/1809.07454)
-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][tasnet]
-repository and added support for multiple audio channels.
+`sdk_version` : _string_  
+Only applicable for `streamlit` SDK.  
+See [doc](https://hf.co/docs/hub/spaces) for more info on supported versions.
 
+`app_file`: _string_  
+Path to your main application file (which contains either `gradio` or `streamlit` Python code).  
+Path is relative to the root of the repository.
 
-Demucs achieves a state-of-the-art SDR performance of 6.3 when trained only on MusDB.
-Conv-Tasnet achieves an SDR of 5.7, to be compared with the best performing spectrogram domain model [D3Net][d3net]
-with an average SDR of 6.
-Unlike Conv-Tasnet, Demucs reacts positively to pitch/tempo shift augmentation (+0.5 SDR). However, Demucs
-still suffers from leakage from other sources, in particular between the vocals and other sources, which is less of a problem
-for Conv-Tasnet. When trained with 150 extra tracks, Demucs reaches an SDR of 6.8, and even surpasses the IRM oracle
-for the bass source (7.6 against 7.1 for the oracle).
-See [our paper][demucs_arxiv] Section 6 for more details or listen to our
-[audio samples][audio] .
-
-<p align="center">
-<img src="./demucs.png" alt="Schema representing the structure of Demucs,
-    with a convolutional encoder, a BiLSTM, and a decoder based on transposed convolutions."
-width="800px"></p>
-
-
-## Important news if you are already using Demucs
-
-See the [release notes](./docs/release.md) for more details.
-
-- 11/05/2021: Adding support for MusDB-HQ and arbitrary wav set, for the MDX challenge. For more information
-on joining the challenge with Demucs see [the Demucs MDX instructions](docs/mdx.md)
-- 28/04/2021: **Demucs v2**, with extra augmentation and DiffQ based quantization.
-  **EVERYTHING WILL BREAK**, please restart from scratch following the instructions hereafter.
-  This version also adds overlap between prediction frames, with linear transition from one to the next,
-  which should prevent sudden changes at frame boundaries. Also, Demucs is now on PyPI, so for separation
-  only, installation is as easy as `pip install demucs` :)
-- 13/04/2020: **Demucs released under MIT**: We are happy to release Demucs under the MIT licence.
-    We hope that this will broaden the impact of this research to new applications.
-
-
-## Comparison with other models
-
-An audio comparison of Demucs and Conv-Tasnet with other state-of-the-art methods such as [Wave-U-Net][waveunet], [OpenUnmix][openunmix] or
-[MMDenseLSTM][mmdenselstm] is available on [the audio comparison page][audio].
-We provide hereafter a summary of the different metrics presented in the paper.
-You can also compare [Spleeter][spleeter], Open-Unmix, Demucs and Conv-Tasnet on one of my favorite
-songs on our [soundcloud playlist][soundcloud].
-
-### 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][demucs_arxiv], Section 5 and 6,
-for more details.
-
-| Model         | Domain     | Extra data?  | Overall SDR | MOS Quality | MOS Contamination |
-| ------------- |-------------| -----:|------:|----:|----:|
-| [Open-Unmix][openunmix]      | spectrogram | no | 5.3 | 3.0 | 3.3 |
-| [D3Net][d3net]  | spectrogram | no | 6.0 | - | - |
-| [Wave-U-Net][waveunet]      | waveform | no | 3.2 | - | - |
-| Demucs (this)      | waveform | no | **6.3** | **3.2** | 3.3 |
-| Conv-Tasnet (this)     | waveform | no | 5.7 | 2.9 | **3.4** |
-| Demucs  (this)    | waveform | 150 songs | **6.8** | - | - |
-| Conv-Tasnet  (this)    | waveform | 150 songs | 6.3 | - | - |
-| [MMDenseLSTM][mmdenselstm]      | spectrogram | 804 songs | 6.0 | - | - |
-| [D3Net][d3net]  | spectrogram | 1.5k songs | 6.7 | - | - |
-| [Spleeter][spleeter]  | spectrogram | 25k songs | 5.9 | - | - |
-
-
-
-## Requirements
-
-You will need at least Python 3.7. See `requirements.txt` for requirements for separation only,
-and `environment-[cpu|cuda].yml` if you want to train a new model.
-
-### For Windows users
-
-Everytime you see `python3`, replace it with `python.exe`. You should always run commands from the
-Anaconda console.
-
-### For musicians
-
-If you just want to use Demucs to separate tracks, you can install it with
-
-    python3 -m pip -U install demucs
-
-Advanced OS support are provided on the following page, **you must read the page for your OS before posting an issues**:
-- **If you are using Windows:** [Windows support](docs/windows.md).
-- **If you are using MAC OS X:** [Mac OS X support](docs/mac.md).
-- **If you are using Linux:** [Linux support](docs/linux.md).
-
-### For machine learning scientists
-
-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
-    pip install -e .
-
-This will create a `demucs` environment with all the dependencies installed.
-
-
-You will also need to install [soundstretch/soundtouch](https://www.surina.net/soundtouch/soundstretch.html): on Mac OSX you can do `brew install sound-touch`,
-and on Ubuntu `sudo apt-get install soundstretch`. This is used for the
-pitch/tempo augmentation.
-
-### Running in Docker
-
-Thanks to @xserrat, there is now a Docker image definition ready for using Demucs. This can ensure all libraries are correctly installed without interfering with the host OS. See his repo [Docker Facebook Demucs](https://github.com/xserrat/docker-facebook-demucs) for more information.
-
-
-### Running from Colab
-
-I made a Colab to easily separate track with Demucs. Note that
-transfer speeds with Colab are a bit slow for large media files,
-but it will allow you to use Demucs without installing anything.
-
-[Demucs on Google Colab](https://colab.research.google.com/drive/1jCegIzLIuqqcM85uVs3WCeAJiSoYq3oh?usp=sharing)
-
-## Separating tracks
-
-In order to try Demucs or Conv-Tasnet on your tracks, simply run from the root of this repository
-
-```bash
-python3 -m demucs.separate PATH_TO_AUDIO_FILE_1 [PATH_TO_AUDIO_FILE_2 ...] # for Demucs
-python3 -m demucs.separate --mp3 PATH_TO_AUDIO_FILE_1 --mp3-bitrate BITRATE # output files saved as MP3
-python3 -m demucs.separate -n tasnet PATH_TO_AUDIO_FILE_1 ... # for Conv-Tasnet
-```
-
-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.
-
-Separated tracks are stored in the `separated/MODEL_NAME/TRACK_NAME` folder. There you will find four stereo wav files sampled at 44.1 kHz: `drums.wav`, `bass.wav`,
-`other.wav`, `vocals.wav` (or `.mp3` if you used the `--mp3` option).
-
-All audio formats supported by `torchaudio` can be processed (i.e. wav, mp3, flac, ogg/vorbis etc.).
-Audio is resampled on the fly if necessary.
-The output will be a wave file, either in int16 format or float32 (if `--float32` is passed).
-You can pass `--mp3` to save as mp3 instead, and set the bitrate with `--mp3-bitrate` (default is 320kbps).
-
-Other pre-trained models can be selected with the `-n` flag.
-The list of pre-trained models is:
-- `demucs`: Demucs trained on MusDB,
-- `demucs_quantized`: Quantized Demucs with [diffq](https://github.com/facebookresearch/diffq),
-    this is much smaller (150MB instead of 1GB) and quality should be exactly the same. Let me know if you disagree.
-    As a result, this is the one used by default.
-- `demucs_extra`: Demucs trained with extra training data,
-- `demucs48_hq`: Demucs with 48 initial hidden channels, trained on [MusDB-HQ](https://zenodo.org/record/3338373),
- used as a baseline for the [Music Demixing Challenge 2021](https://www.aicrowd.com/challenges/music-demixing-challenge-ismir-2021),
-- `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 the *shift trick*) 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 but it does make vocals a bit smoother.
-
-The `--overlap` option controls the amount of overlap between prediction windows (for Demucs one window is 10 seconds).
-Default is 0.25 (i.e. 25%) which is probably fine.
-
-
-### 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! 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
-```bash
-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][musdb].
-To start training on a single GPU or CPU, use:
-```bash
-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:
-```bash
-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`.
-
-
-### MusDB HQ
-
-To train on MusDB HQ, use the following flags:
-
-```bash
-python3 -m demucs -b 4 --musdb MUSDB_HQ_PATH --is_wav [...]
-```
-
-### Custom wav dataset
-
-You can trained on a custom wav dataset using the following command.
-At the moment, you still need to pass the MusDB path for evaluation, and the model
-must use the standard sources (bass, drums, other, vocals). However, it should be relatively
-easy to fork the code to support different patterns.
-
-```bash
-python3 -m demucs -b 4 --wav PATH_TO_WAV_DATASET [...]
-```
-
-The folder `PATH_TO_WAV_DATASET` should contain two sub-directories : `train` and `valid`. Each of those
-should contain one folder per track. Each track folder must contain one file for each source (`drums.wav`, `bass.wav`, `other.wav`, `vocals.wav`) and one file for the mixture (`mixture.wav`).
-
-By default, the custom wav dataset will replace MusDB. To concatenate it with MusDB, pass `--concat` (if you are using musdbhq, dont forget to pass `--is_wav`).
-
-### Fine tuning
-
-You can fine tune from one of the pre-trained models listed in the [Separating tracks Section](#separating-tracks)
-by passing the `--init=PRETRAINED_NAME`, i.e. for Demucs or ConvTasnet:
-
-```bash
-python3 -m demucs -b 4  --musdb MUSDB_PATH --init demucs # Demucs
-python3 -m demucs -b 4  --musdb MUSDB_PATH --tasnet --samples=80000 --split_valid --init tasnet # Conv-Tasnet
-```
-
-### 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](demucs/parser.py) for more details.
-
-
-### Test set evaluations
-
-Test set evaluations computed with [museval][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:
-
-```bash
-python3 -m demucs.raw [--workers=10] MUSDB_PATH RAW_PATH
-```
-
-You can then train using the `--raw RAW_PATH` flag, for instance:
-```bash
-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:
-```bash
-# 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
-```bash
-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](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:
-```bash
-export DEMUCS_MUSDB=PATH TO MUSDB
-# Optionally, if you extracted raw pcm data
-# export DEMUCS_RAW=PATH TO RAW PCM
-```
-
-## How to cite
-
-```
-@article{defossez2019music,
-  title={Music Source Separation in the Waveform Domain},
-  author={D{\'e}fossez, Alexandre and Usunier, Nicolas and Bottou, L{\'e}on and Bach, Francis},
-  journal={arXiv preprint arXiv:1911.13254},
-  year={2019}
-}
-```
-
-## License
-
-Demucs is released under the MIT license as found in the [LICENSE](LICENSE) file.
-
-The file `demucs/tasnet.py` is adapted from the [kaituoxu/Conv-TasNet][tasnet] repository.
-It was originally released under the MIT License updated to support multiple audio channels.
-
-
-[nsynth]: https://magenta.tensorflow.org/datasets/nsynth
-[sing_nips]: https://research.fb.com/publications/sing-symbol-to-instrument-neural-generator
-[sing]: https://github.com/facebookresearch/SING
-[waveunet]: https://github.com/f90/Wave-U-Net
-[musdb]: https://sigsep.github.io/datasets/musdb.html
-[museval]: https://github.com/sigsep/sigsep-mus-eval/
-[openunmix]: https://github.com/sigsep/open-unmix-pytorch
-[mmdenselstm]: https://arxiv.org/abs/1805.02410
-[demucs_arxiv]: https://hal.archives-ouvertes.fr/hal-02379796/document
-[musevalpth]: museval_torch.py
-[tasnet]: https://github.com/kaituoxu/Conv-TasNet
-[audio]: https://ai.honu.io/papers/demucs/index.html
-[spleeter]: https://github.com/deezer/spleeter
-[soundcloud]: https://soundcloud.com/voyageri/sets/source-separation-in-the-waveform-domain
-[original_demucs]: https://github.com/facebookresearch/demucs/tree/dcee007a350467abc3295dfe267034460f9ffa4e
-[diffq]: https://github.com/facebookresearch/diffq
-[d3net]: https://arxiv.org/abs/2010.01733
+`pinned`: _boolean_  
+Whether the Space stays on top of your list.

app.py

@@ -0,0 +1,26 @@
+import os
+import gradio as gr
+from scipy.io.wavfile import write
+
+
+def inference(audio):
+  os.makedirs("out", exist_ok=True)
+  write('test.wav', audio[0], audio[1])
+  os.system("python3 -m demucs.separate -n mdx_extra_q -d cpu test.wav -o out")
+  return "./out/mdx_extra_q/test/vocals.wav","./out/mdx_extra_q/test/bass.wav",\
+"./out/mdx_extra_q/test/drums.wav","./out/mdx_extra_q/test/other.wav"
+  
+title = "Demucs"
+description = "Gradio demo for Demucs: Music Source Separation in the Waveform Domain. To use it, simply upload your audio, or click one of the examples to load them. Read more at the links below."
+article = "<p style='text-align: center'><a href='https://arxiv.org/abs/1911.13254' target='_blank'>Music Source Separation in the Waveform Domain</a> | <a href='https://github.com/facebookresearch/demucs' target='_blank'>Github Repo</a></p>"
+
+examples=[['test.mp3']]
+gr.Interface(
+    inference, 
+    gr.inputs.Audio(type="numpy", label="Input"), 
+    [gr.outputs.Audio(type="file", label="Vocals"),gr.outputs.Audio(type="file", label="Bass"),gr.outputs.Audio(type="file", label="Drums"),gr.outputs.Audio(type="file", label="Other")],
+    title=title,
+    description=description,
+    article=article,
+    examples=examples
+    ).launch(enable_queue=True)

baselines/IRM2/test/AM Contra - Heart Peripheral.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7a1e79ff415009526e480beba3666e14b163fe33c23dab4040e2077e25c61bbe
-size 26828

baselines/IRM2/test/Al James - Schoolboy Facination.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:8caae5351b7dd0bc7fce2ee4cc915e291ed4636709dfbaf19962aee0f0a618ab
-size 24865

baselines/IRM2/test/Angels In Amplifiers - I'm Alright.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ed9d8a341ceb370fa8e8b7cc6f336def7efa00941edac1cb834f573fd5c82253
-size 23052

baselines/IRM2/test/Arise - Run Run Run.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9066cfe82dc69614fae9338dcfdef5a65f7e2ebdb7dadadb6d27df321d6c3005
-size 25900

baselines/IRM2/test/BKS - Bulldozer.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9a54943bbd84ca3474a1bf2985d5c34796b1eebe8291f9d84807b1b81695cf24
-size 42911

baselines/IRM2/test/BKS - Too Much.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:57c5c90e3cbb955ce0210cc6d5aaad3af84ff23493aea10e6c30f189fc364c87
-size 21123

baselines/IRM2/test/Ben Carrigan - We'll Talk About It All Tonight.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ed5db465dd93085a257b3044c55c2381ffd60845067ed47f0d6ad6d9b1061d10
-size 20579

baselines/IRM2/test/Bobby Nobody - Stitch Up.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:74f77a2bf52249b78c23a9be577de79d9168d95cf00fb3a674468ec35342f5ba
-size 23353

baselines/IRM2/test/Buitraker - Revo X.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:feee80b8fa70d36eb2ef8a7de45c087db7cec4b6b0145218d14a4be99ed9f785
-size 28203

baselines/IRM2/test/Carlos Gonzalez - A Place For Us.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3aaadf03103b92715015653ec7d5af59db586dc8d45c4ced2b2ef5e4db4f8e83
-size 31961

baselines/IRM2/test/Cristina Vane - So Easy.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:8470eebddeb221bf2702c162cc7f19c2d5e7f8eb14d6520ad1e060165c7773b6
-size 28275

baselines/IRM2/test/Detsky Sad - Walkie Talkie.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:bf8778f15c4ad7bb3fa30428ca2d780be40b22f3268227d01e5ae803afda180c
-size 11024

baselines/IRM2/test/Enda Reilly - Cur An Long Ag Seol.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:14e7d069a2840224a8876fdc7f328ca52a865e6540a2939c3c2aef0ccd6fe7df
-size 21139

baselines/IRM2/test/Forkupines - Semantics.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ba47a499b526cde808fbb10ca2858886c4606f1bccee921ae54721091573a80f
-size 21237

baselines/IRM2/test/Georgia Wonder - Siren.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:cca85c851447e94539bfdefe0b00a6fcb4746e34197058b2dc8b4feb46b193a1
-size 36791

baselines/IRM2/test/Girls Under Glass - We Feel Alright.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9a7e24e0628f44ee1cc774e2c454ac4cea0a9c63064348af6e33225505c2f1d2
-size 21935

baselines/IRM2/test/Hollow Ground - Ill Fate.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:8f37c5bc36becaed3d9649b36e6d68b8f4a86960b4103fcc2f5710fa772dae37
-size 12330

baselines/IRM2/test/James Elder & Mark M Thompson - The English Actor.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:09faaffbfd9825a3104c935ac27da722700c895f1c5c095b6f9570ed551a189a
-size 20687

baselines/IRM2/test/Juliet's Rescue - Heartbeats.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a5ec942c2019fabba64dbbcad131b7ac986f126d87fdb9ebbd3c8d139217d158
-size 27498

baselines/IRM2/test/Little Chicago's Finest - My Own.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:fdaf06d216ad224f0c78ca9163f190311157718618e17e462bd781d1b36c3e25
-size 31610

baselines/IRM2/test/Louis Cressy Band - Good Time.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f7191c053ebc148d9ca86e6583cb0ecafaa773e4e0107a5d7c0915bab034c03e
-size 21264

baselines/IRM2/test/Lyndsey Ollard - Catching Up.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a54bda98c9b061c4c994b173df4b1c895de56475484cb29973e2a86c5133699b
-size 25709

baselines/IRM2/test/M.E.R.C. Music - Knockout.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1e81ed157dec77de057bbb712ad78bf574305d8d2ef4d24e84f2a8ca69ce88ee
-size 26605

baselines/IRM2/test/Moosmusic - Big Dummy Shake.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:164e3f93976f698750757c21dd5c77e932c90b7f5cbccae6c44a976abc86ff86
-size 22563

baselines/IRM2/test/Motor Tapes - Shore.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1938098d0212e9afdff2a0371887b66842718fdca141ce9c017ceadc9e2ef822
-size 25089

baselines/IRM2/test/Mu - Too Bright.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:68ca54ac8985fbad1c47137ced2e302831bb9c5b35ddb2c6c45ec4f8848ea3bf
-size 22522

baselines/IRM2/test/Nerve 9 - Pray For The Rain.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:829f8efcda9469b277961892dff22d4b569a3e57f7cc29bc773e686d9a23455b
-size 32881

baselines/IRM2/test/PR - Happy Daze.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:e0d41b2eb0aff25e1e1ba1e884a50257bfb1e87905bc396a2cd2cf6eda0f5e7d
-size 21052

baselines/IRM2/test/PR - Oh No.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:cd6a60a64165c2f4f427f041c6955a18776858201afc293c9eb50f92b82369bf
-size 9804

baselines/IRM2/test/Punkdisco - Oral Hygiene.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:4186d9e9b295ffc6ef41e39534d1109d19b0182cb92288b7a7abdca8afb1aaff
-size 19114

baselines/IRM2/test/Raft Monk - Tiring.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:535a886ce8cdf756f489fc36cf7d73298d09839e241121f5039046c96be1d20b
-size 23263

baselines/IRM2/test/Sambasevam Shanmugam - Kaathaadi.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:648a338396f76e002c61f3bd10323cebc155a7675e630ac6dbee8335fcdf2828
-size 23183

baselines/IRM2/test/Secretariat - Borderline.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ac1a105faf4ee9bb3ec09273d21aeadf5de7752cb0eb0ca292f673210d77bfd3
-size 27299

baselines/IRM2/test/Secretariat - Over The Top.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2e1a432abaa8b5e5bf6fdc6df892706e047dc5bd96e2e2da874dca5f14684441
-size 21642

baselines/IRM2/test/Side Effects Project - Sing With Me.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:594dc701ab7ce759ff1006e8947688f1609e9bc87a18aba638b83432d9e2d32e
-size 28539

baselines/IRM2/test/Signe Jakobsen - What Have You Done To Me.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c9347bd26e385f06a0f7350534a7f205245765e6d3eb145922cdb4cf3de4f0f7
-size 22512

baselines/IRM2/test/Skelpolu - Resurrection.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:494b2d48e6d33b2167f542cc1b63c45377514f126245fd634d270f34495f8251
-size 14837

baselines/IRM2/test/Speak Softly - Broken Man.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:8e049a84e184aed110458ca26ac929ddc79caef637dfa8decdcb4d438afb4232
-size 25457

baselines/IRM2/test/Speak Softly - Like Horses.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a29311792a22e9e15c970d8e9a84932e54ed969d0efc2dcd8ba188e409924a08
-size 27657

baselines/IRM2/test/The Doppler Shift - Atrophy.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3beaba378e7e0c92d1204fcd7712aeecc09c8ee63c3e56b45e9357d64ae426a1
-size 42460

baselines/IRM2/test/The Easton Ellises (Baumi) - SDRNR.json.gz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d6b5f83f1984d9440a4e5ceb5c68d80dc84d887861da24a9dc96017aa2a399f5
-size 29888