Logo Music Information Retrieval at Vienna IFS
Vienna University of Technology
Institute of Software Technology and Interactive Systems
Information & Software Engineering Group

Music Information Retrieval

TU Logo  IFS Logo
  [Projects] [Downloads] [People] [Publications] [Press] [Events]  

Audio Feature Extraction Webservice

Self-Organising (Music) Map Webservice

Webservice Client

Webservice Interface

IFS mir group Webservices

On this web page we provide web services for

  • automatic music audio feature extraction over the web
  • training of self-organizing music maps

After an introduction to what the web services exactly can do, a demo webservice client is provided and also the specifications which allow the use of this web service in your own applications.

Audio Feature Extraction Webservice

Our webservice enables the automatic extraction of semantic information from music (wav or mp3 files) over the Internet.

By using methods from digital signal processing and psycho-acoustics, the features extracted from the audio signal provide a numeric description of the acoustic content of the music, providing information about rhythm, timbre, etc. More specificially, three feature sets - Rhythm Patterns, Statistical Spectrum Descriptors and Rhythm Histograms - are extracted [more details], and returned as a space separated ASCII file in SOMLib format.


With the use of these features it is possible to tell about the content of a piece of music without the need of annotated labels such as artist, song title or genre. With this information, in turn, similarity between pieces of music can be computed and applications such as music classification, automatic music organization, playlist recommendation and many more are made possible.

Webservice for training Self-Organising (Music) Maps (not available anymore)

Our group also conducts research on data mining methods and in particular the Self-Organizing Map (SOM). The SOM is a prominent technique for exploratory data analysis and visualization of complex high-dimensional data sets [more about SOMs].

We have successfully introduced a SOM application, that organizes entire music collections on a Self-Organizing Music Map providing convenient overview and immediate interaction with your music collection, organized by music similarity - including creation of situation/style-based playlists. [more about PlaySOM and music maps]

We also provide a web service for the training and creation of Music SOMs (however, not yet our viewer application).

Webservice Client

If you want to play with our feature extractor or the SOM-trainer we provide a simple client implemented in Java. After you received your voucher you can easily launch the client using Java Web Start. This client requires Java1.6.

Because the client needs access to your computer's storage and network connection the jar-file is signed. Java Web Start will prompt a warning that "The application's digital signature cannot be verified." This is normal, so just click "Run" to finally start the Demo-Client. If you want to check the certificate, you can compare the MD5 Fingerprint: 83:DE:BC:30:D6:7B:7B:3F:22:93:E5:A6:53:D8:94:1E

Using this client you can easily analyse your mp3-files and afterwards create a SOM containing your music. Please keep in mind that all files you are going to analyze will be sent to our webservice over the Internet, so your traffic-counter might increase quite fast...

Webservice Interface

We have created webservices that allow online audio feature extraction as well as creation and training of self organizing maps. You are welcome to use them - just get the right wsdl-file and build you personal client:

Also for using the Webservice with your own client you need to get a voucher which has to be sent along with every requests. You will get a free voucher valid for 30 days. Request your voucher here. (The voucher registration is currently not available)

For more detailed information on the Webservice-interface please look at the javadoc-API.


Part of this work was supported by the European Union in the 6th Framework Program, IST, through the MUSCLE NoE on Multimedia Understanding through Semantics, Computation and Learning, contract 507752.

last edited 23.03.2015 by Thomas Lidy