task-speechopen_acq-pangolin_events.json

{
  "onset": {
    "Description": "Time elapsed from the start of EEG recording. For overt/listening events, computed by aligning the VAD-detected speech onset in the audio file to the EEG timeline using the estimated recording delay (estim_delay.json). For covert events, directly derived from trigger-based EEG sample indices at 1200 Hz.",
    "Units": "s"
  },
  "duration": {
    "Description": "Duration of the event. For overt/listening events, this is the VAD-detected speech segment duration excluding 0.1 s padding. For covert events, this is fixed at 5.0 s (experimental design).",
    "Units": "s"
  },
  "trial_type": {
    "Description": "Type of speech event.",
    "Levels": {
      "overt": "Overt (aloud) speech production by the participant.",
      "speech_window": "Fixed 5 s analysis window extracted from overt speech production.",
      "listening": "Listening to speech stimuli presented via audio playback.",
      "covert": "Covert (silent/imagined) speech production by the participant, following a listening stimulus."
    }
  },
  "wav_onset": {
    "Description": "Onset time relative to the start of the audio recording file (before EEG-audio delay correction). This is the unpadded speech onset in the WAV file timeline. Set to 'n/a' for covert events which have no associated audio.",
    "Units": "s"
  },
  "sample": {
    "Description": "EEG sample index corresponding to the event onset (onset * SamplingFrequency, where SamplingFrequency = 1200 Hz).",
    "Units": "samples"
  },
  "value": {
    "Description": "Speech transcript generated by ASR (kotoba-whisper-v2.0-faster). 'n/a' indicates the transcript was unavailable."
  }
}