Data and code are on scratch. See this page for details on access: https://neuromorphs.net/nm/wiki/2012/NetworkServices

The code is in s:att12\Ed Code

The first data files look like


the first part is always att12, the second part is the date the data was collected.

There are 6 playlists preprocessed in different forms: enhanced, m-sequence, original, m-sequence, original, enhanced

The first three file formats come out of pyCorder:

Ed processes this data, and creates

  • raw - Mat file, unfiltered, stored by Ed

He then filters and cleans up the data using functions from EEGLab

  • preprocessed - Filtered between 1 and 50Hz. Vector called onsets indicates the starting sample of each of the ten 1-minute segments.
  • _BP_SSP - cleaned up by sid for second day's experiments. (Downsampled is for Sid and downsampled to 100Hz).

The preprocessed file contains the entire EEG signal for the experiment, which consists of 10 trials each 1 minute long, and is sampled at 1000Hz. The onset variable shows the start of the clicks. Advance by 3000 samples to get to the start of the actual speech signal.

First day's data: we only got 31 channels. Other days we got 32 channels.

The first day's data (one speaker) is in: s:att12\eeg_data_70911_sudar

The second experiment (two speakers) is in: s:att12\two_stimuli_eeg, with the stimuli in s:att12\two_stimuli. Second day, we only looked at the original speech (since the enhanced and m-sequence speech didn't help that much.)

Subjects already listen to the Journey to the center of the earth story. The stimuli are labeled as follows:

  • diotic - both speech signals are identically presented to both ears.
  • dichotic_L - Each ear hears only one signal, attend to Journey in the left
  • HRTF_L - Spatialized sounds, with attention directed to Journey in the left