ffrprep.analysis.stim_to_resp_xcorr

ffrprep.analysis.stim_to_resp_xcorr(stim, resp, sfreq, lag_range_ms=None)[source]

Maximum stimulus-to-response cross-correlation (coefficient form).

Both inputs must share the sampling rate sfreq and be already cropped to the analysis windows (e.g. stimulus vowel 50-170 ms vs. response 60-180 ms). The cross-correlation is normalized like MATLAB’s xcorr(..., 'coeff'): divided by sqrt(sum(stim**2) * sum(resp**2)) with no mean removal. The reported value is the maximum over lags (optionally restricted to lag_range_ms).

Parameters:

  • stim (array-like) – 1-D stimulus and response waveforms sampled at sfreq.

  • resp (array-like) – 1-D stimulus and response waveforms sampled at sfreq.

  • sfreq (float) – Sampling frequency in Hz.

  • lag_range_ms (tuple of (float, float), optional) – Inclusive lag window in ms in which to search for the maximum. Positive lag means the response is delayed relative to the stimulus. None searches all lags.

Returns:

  • r (float) – Maximum correlation coefficient within the searched lags.

  • z (float) – Fisher r-to-z transform of r (arctanh).

  • lag_ms (float) – Lag in ms at which the maximum occurs.