Skip to content

Abstract Stimulus Generation Method

This is an abstract class describing all features common to a stimulus generation method. In addition to these features, the generate_stimulus method is common to every stimulus generation type. Furthermore, this abstract class contains properties common to all stimulus generation methods.

Abstract Properties

These are automatically instantiated for subclasses, since they are not abstract themselves. Default values are shown here:

- min_freq = 100 % The minimum frequency a stimulus can have (Hz)
- max_freq = 22e3 % The maximum frequency a stimulus can have (Hz)
- duration = 0.5 % The duration of each stimulus (sec)
- n_trials = 100 % The number of trials to generate
- Fs = 44.1e3 % The sampling rate (Hz)

subject_selection_process

[y, spect, binned_repr] = subject_selection_process(self, signal)

Model of a subject performing the task. Takes in a signal (the gold standard) and returns a self.n_trials x 1 vector of -1 for "no" and 1 for "yes".

generate_stimuli_matrix

[stimuli_matrix, Fs, spect_matrix, binned_repr_matrix, W] = generate_stimuli_matrix(self)

Generates a matrix of stimuli. Explicitly calls the generate_stimulus() class method.

OUTPUTS:

  • stimuli_matrix: n x self.n_trials numerical vector, the stimulus waveform, where n is self.nfft + 1.
  • Fs: 1x1 numerical scalar, the sample rate in Hz.
  • spect_matrix: m x self.n_trials numerical vector, the half-spectrum, where m is self.nfft / 2, in dB.
  • binned_repr_matrix: self.n_bins x self.n_trials numerical vector, the binned representation.
  • W: p x self.n_trials or [], where p is the size of the weight matrix. Only full if self is HierarchicalGaussianStimulusGeneration.

See Also

from_config

Set properties from a struct holding config options.

See Also

synthesize_audio

Synthesize audio from spectrum, X. If X is an array, each column is treated as a spectrum.