Data collected for experiments with eyetracking during phonetic segmentation tasks.

See the releases to download the data.

The data consists of a YAML file containing a list of static scenes viewed by each subject ("vp"), with associated gaze events and relevant UI details. Example excerpt:

- start: 2014-12-10T08:24:16.033Z
  end: 2014-12-10T08:24:18.033Z
  windowStart: 0.0
  windowEnd: 3.3231078705735193
  gaze:
  - vp: '02'
    timeStamp: 2014-12-10T08:24:16.416Z
    signalTime: 1.2743769
    gazeType: Fixation
    gazeDur: 425.0
    gazeRegion: spectrogram
    position:
      xPos: 761
      yPos: 579
    subRegion: middle
  - vp: '02'
    timeStamp: 2014-12-10T08:24:16.858Z
    signalTime: 1.3288541
    gazeType: Fixation
    gazeDur: 175.0
    gazeRegion: spectrogram
    position:
      xPos: 789
      yPos: 563
    subRegion: top

For each subject, we provide a video in Matroska format (which can be played in VLC and other free software), including

• screen capture of the session (video stream #1)
• audio played back during the segmentation task
• rendered acoustic waveform and spectrogram for each scene, and gaze fixations (video stream #2)
• subtitles indicating start and end time of each scene in the segmented recording

The second video stream has been reconstructed from the segmented audio, structured scene information, and gaze data (to be released).

For details, see the paper presented at LREC 2018:

@inproceedings{Khan2018LREC,
  year = {2018},
  month = may,
  booktitle = {11th Language Resources and Evaluation Conference (LREC)},
  address = {Miyazaki, Japan},
  author = {Khan, Arif and Steiner, Ingmar and Sugano, Yusuke and Bulling, Andreas and Macdonald, Ross},
  date = {2018-05-11},
  eprint = {1712.04798},
  eprinttype = {arxiv},
  pages = {4277-4281},
  title = {A Multimodal Corpus of Expert Gaze and Behavior during Phonetic Segmentation Tasks},
  url = {http://www.lrec-conf.org/proceedings/lrec2018/summaries/676.html}
}