Skip to content. | Skip to navigation

Personal tools


You are here: Home / Codes, API & data resources

Codes, API & data resources


Open source code:

1. Front-end code of the inEvent portal:

The open source code of the inEvent online portal is available on github, under the MIT license (yielding a free access to the software without restriction, and without limitation to use, copy, publish, and distribute it with an acknowledgement). A quick installation guide is also provided.

2. Speaker diarization code:

3. Hyperevent recommendation code:
Content-Based Recommendation

Emotion-Based Recommendation

4.Transcription producer and meeting summarization code:


API specifications:

1. Main links - DTOdefinitions

the API specifications are available online. The Data Transfer Objects and REST services defined for archiving, retrieving, and processing multimedia are described in the URL above

2. Public dataset retrieval:

The retrieval section of the API documentation, describes how to download all the dataset, or a subset of it based on the filter and search query specified.

Every event retrieved is returned along with:

  • Its metadata
  • Related events (based on emotion-based and content-based similarity).
  • Tracks (video files, slides, audio and video processing outputs when available.)


3. Access to Audio and Video Processing Services via the REST API:

The processing section of the API documentation describes the specifications defined inEvent and adopted to issue a Web service processing request for:

  • Speaker diarization.
  • Automatic speech recognition.
  • And video analytics.


Public dataset description:

The public dataset covers 1250 hyperevents representing approximately 1500 hours from Klewel, Radvision, and TED talks along with their metadata, and the output of audio and visual analytics (for a subset of them), in the form of tracks.

Radvision’s dataset contains a total of 14 recordings of inEvent meetings, and inEvent Webinar 2013 and Fresh 2014. A detailed description of the TED dataset used is available here:

Klewel’s dataset contains university lecture and events such as international congresses, conferences, and symposiums recorded using Klewel’s capture station. Klewel’s dataset consists of 123 hyperevents, corresponding to approximately 73 recorded hours of content including EPFL lectures and presentations from the Djangocon 2012, Fresh2012, and Lift 2011 conferences.

EU Project

Networked Media and Search Systems

European flag

This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no 287872”.