Contribute a new dataset
To contribute a new dataset, please
- create an account
(you will need to enter the code lva2012 to avoid spamming) - login and go to the wiki page below corresponding to the nature of your data (synthetic, audio, biomedical, other)
- click on the Edit link at the bottom of the page and add a new section for your dataset (the datasets are ordered according to the year of collection)
- click on the Save link at the bottom of the page to save your modifications
Please make sure to provide at least the following information:
- name of the dataset and year of collection
- contact information (name, institution, email address or web page)
- link to the dataset
- short description of the dataset (nature of the data, number of sensors and samples, license, etc)
If possible, please upload the dataset at a stable URL on the website of your institution and provide its URL only. If not, you may attach the dataset to the wiki page as a single .ZIP or .TGZ file instead.
Evaluation Campaigns
A number of large-scale datasets for audio and biomedical signal separation are available from the websites of the following evaluation campaigns:
- 2nd CHiME Speech Separation and Recognition Challenge (2013)
- 3rd Signal Separation Evaluation Campaign (2011)
- 1st CHiME Speech Separation and Recognition Challenge (2011)
- 2nd Signal Separation Evaluation Campaign (2010)
- 1st Signal Separation Evaluation Campaign (2008)
- 1st Stereo Audio Source Separation Evaluation Campaign (2007)
- MLSP 2007 Data Analysis Competition
- 1st PASCAL Speech Separation Challenge (2006)
- MLSP 2006 Data Analysis Competition
- MLSP 2005 Data Analysis Competition
Using these datasets will allow you to compare your results to those of many other people.