Freesound Labs

This repository contains code for the frontpage of the Freesound Labs website (https://labs.freesound.org). It uses Jekyll to generate a static blog-like website.

Adding content

To add new posts, papers or datasets, please create Markdown files using the instructions below and open a new pull request with the changes.

Posts

Here is an example template for new posts (which should be placed in the _posts folder):

---
layout: post
title:  "Title of the entry"
date:   YYYY-MM-DD 12:00:00
release_date: YYYY-MM-DD 12:00:00
categories: 
- Apps
- Research
- Educational
tags: 
- freesound-api
- music-creation
project_url: https://...
code: https://...
image: /filename.png
by: 
- Name Surname
- Name Surname
---

Description for the entry.


<iframe width="560" height="315" src="https://www.youtube.com/embed/YOUTUBE_ID" frameborder="0" allowfullscreen></iframe>

Papers

Papers can be automatically added using the retrieve_papers.py script. However, papers can also be manually added by adding their Semantic Scholar URL to the file manual_paper_semantic_scholar_urls.txt (one per line). The retrieve_papers.py script will combine the papers manually introduced with the automatic citations found using Semantic Scholar API. See instructions below for running the script.

Datasets

New dataset entries should be added to the _datasets folder using the following template:

---
layout: post
title: "Dataset name"
date: 2018-03-15
project_url: https://dataset.url/with/description/and/download
image: /optional_image.png
institutions:
- Music Technology Group of Universitat Pompeu Fabra
- Institution 2
authors: 
- Name Surname
- Name Surname
- Name Surname
---
DatasetX is an audio dataset containing 11,073 audio files annotated with 41 labels of the [AudioSet Ontology](https://research.google.com/audioset////////ontology/index.html). FSDKaggle2018 has been used for the [DCASE Challenge 2018 Task 2](http://dcase.community/challenge2018/task-general-purpose-audio-tagging), which was run as a Kaggle competition titled [Freesound General-Purpose Audio Tagging Challenge](https://www.kaggle.com/c/freesound-audio-tagging). The description should not be very long or it will look bad in the website ;)

Development

Freesound Labs website can be build using a Dockerized Jekyll installation which comes with all necessary dependencies bundled in.

First you'll need to build the "builder" image (in case it has not been built already). Simply run:

docker build -t fslabs-builder .  

Then you can build the Freesound Labs website with the command:

docker run --rm --volume="$PWD:/srv/jekyll" -it fslabs-builder jekyll build

You can run the development server with the command:

docker run --rm --volume="$PWD:/srv/jekyll" -p 4000:4000 -it fslabs-builder jekyll serve

...and point your browser to http://0.0.0.0:4000. Changes in will trigger the re-generation of the site.

The retrieve_papers.py script can be run with the following command:

docker run --rm --volume="$PWD:/srv/jekyll" -it fslabs-builder python3 retrieve_papers.py

It should be run before building the site for the papers to appear in it.

NOTE: all commands above should be run from the root of the repository

Deployment

After page updates are commited to this repository, a deployment job needs to be manually run from the mtg-deploy repository (see README there under fs-labs folder) for the changes to become public.

Nota that this job does not automatically update the list of papers of the "papers" section. To that end, the retreive_papers.py script should be run and the new files created should be added and committed to this repo before running the dpeloyment job from `mtg-deploy.

To do that, run the retreive_papers.py script locally using:

docker run --rm --volume="$PWD:/srv/jekyll" -it fslabs-builder python3 retrieve_papers.py