Scripts¶
Housed within this folder are data processing and analysis scripts.
adjust_individual_resynth¶
A simple script to run the adjustment protocol for a given config file
analyze_consistency¶
This script compares the consistency in reconstructions from data collected several days in a row by NB with all settings the same
Setup
analyze_phase2¶
This script contains two different analysis options
for assessing the functionality of a "phase 2" in RC experiments.
Phase 2 consists of taking the original/standard RC reconstruction
and using it as the basis for generating new stimuli.
Those options can be seen in the local function local_create_files_and_stimuli_phaseN
compare_norena¶
Run simulated observer on Norena stimgen and two UniformPrior options (1 bin filled and multiple bins filled) and visualize the results
compare_recons¶
The purpose of this script is to assess the qualitative effect of the peak-sharpening procedure. Runs adjust_resynth.m followed by follow_up.m on select configs/associated data.
compare_stim_and_recon¶
Runs the simulated observer on two UniformPrior options (1 bin filled and multiple bins filled) and uses several different reconstruction methods (ten scale, ridge regression, linear) and visualizes the results
hyperparameter_sweep_custom¶
Hyperparameter Sweep Custom Stimulus
Evaluate hyperparameters of stimulus generation using the 'custom' stimulus paradigm. Evaluate hyperparameters over different target signals.
optimize_hierGauss¶
Run a grid search on HierarchicalGaussian stimgen parameters to determine the best combination of broad, medium, and narrow bases.
optimize_resynth¶
Determine the best mult and binrange parameters for resynthesis. This can be used to extrapolate a good range to present to subjects for a given bin number.
patient_reconstructions¶
Generate reconstructions and visualizatinos for non-target sound data Includes lots of flags for response prediction analysis NOTE: should also work with make_figures_paper2 (not recently tested though)
pilot_reconstructions¶
Compute reconstructions for the pilot data experiment (with target signal). This code assumes that each each experiment uses the same number of bins and that the reconstructions should be done over the bin representation.
OUTPUTS: - T: a data table that contains information about the experiments and their reconstructions
predict_on_TSFdata¶
Run just the cross validation section on data used to train ML model Uses data formatted to be digestable by the TinnitusStimulusFitter package.
reconstruction_viz¶
Visualization for reconstructions
Run pilot_reconstructions.m first to generate recons, etc.
survey_viz¶
This script collects and analyzes subjective rankings from ompare_recons.m
target_signal_sparsity¶
Quantify the sparsity of the target signals (ATA tinnitus examples) in the DCT basis.