SRTF Dataset



  • The neural responses have been measured in the primary auditory cortex (AI) of
    anesthetized mongolian gerbils using an multi-electrode array. Dynamic Moving
    Ripples (DMRs) and complexes of Frequency-Modulated tones (FM banks) have
    been used as stimuli.
    The included responses are recordings from two electrodes (3 units per electrode)
    in the same gerbil for the two stimulus classes.

Dataset format

  • The folder 'DMR_Run_01' and 'FMBanks_Run_01' contain datasets for the different stimulus classes. The files belonging to a specific channel/unit combination can be found in a mapping file. For example, the content of the mapping file '' looks like:

Stimuli/DMR_Run_01_01.wav; Responses/chan01_unit01/DMR_Run_01_01_chan01_unit01.spikes
Stimuli/DMR_Run_01_02.wav; Responses/chan01_unit01/DMR_Run_01_02_chan01_unit01.spikes

  • where each row corresponds to one stimulus and stimulus file and response file locations are separated by a semicolon. There are Matlab routines available for reading a dataset from a mapping file (see
  • Stimulus format:
    The stimuli are given as wav-files (44100Hz, 16 bits)
  • Response format:
    The first line of a response file (*.spike) contains the AD sampling frequency in Hz and the second line the number of trials. All subsequent lines contain the spike times, one line per trial. Hence, a typical spike file looks like:

0.0827499999999873 0.0932750000000169  ...
0.0716999999999189 0.0941249999999172  ...


  • The data has been used in the following publications:

[1] Max F. K. Happel, Simon Müller, Jörn Anemüller, and Frank W. Ohl. Predictability of strfs in auditory cortex neurons depends on stimulus class. In INTERSPEECH-2008, page 670, 2008.

[2] Arne F. Meyer, Max F. K. Happel, Frank W. Ohl, and Jörn Anemüller. Estimation of spectro-temporal receptive fields based on linear support vector machine classification. BMC Neuroscience,
10(Suppl 1):P147, 2009.