Speech In Noise (SIN) Software Package

Overview

SIN is a software package that has been used to run many behavioral tests, including basic psychoacoustic and highly demanding audiological paradigms, at multiple testing sites and Univerisities. For example, SIN has been used to do everything from basic playback and recording to administering audiovisual clinical tests like the Multimodal Lexical Sentence Test (MLST) developed by Karen Kirk and colleagues at the University of Iowa (http://www.flintbox.com/public/project/23463).

The package is written primarily in MATLAB and leverages PsychToolBox (http://psychtoolbox.org/) to create a uniform testing environment with virtually any hardware setup.

Why Use SIN?

Motivation and Development Goals

SIN was initially developed for a multi-site project between the University of Washington and University of Iowa. The project required experimenters to administer approximately 30 - 40 individual audio or audiovisual tasks in a highly reproducible way with minimal inter-experimenter variance across 250 patients. While there are procedures and, in rare instances, applications to administer individual tests, many of these procedures require the experimenter to make real-time decisions and manual adjustments; thus, mistakes are likely, difficult to recover from, and, more importantly, often go unnoticed. In our specific application, it would approximately 500 unique decision points for each participant. That's 500 potential mistakes that would be untraceable and uncorrectable using established procedures.

Thus, the goal SIN was to create a flexible and uniform platform upon which all tests could be administered, scored, and stored with minimal experimenter input.

Design

Front End

As mentioned above, one of my primary goals in writing SIN was to create something that is easy to use for experimenters of all shapes, sizes, ages, and abilities. While I generally prefer command-line work and batch scripting for analyses, this specific application required an easy-to-use GUI based front end.

For most experimenters, runSIN.m is the only function you'll ever need to invoke explicitly. This will launch a simple GUI outfitted with all the functionality you need to register new subjects, run tests, explore results, run analyses, etc.

A detailed walkthrough of the GUI is forthcoming. For immediate questions, please e-mail me directly as this will help shape the walkthrough.

Under the Hood

SIN is comprised of four major types of functions:

1. Players

• A player generally accepts a set of stimuli and parameters to configure playback, recording, modification checks, modifiers, etc.

2. Modification Checks

• Modification checks are generally used to gather some form of information and convert that information into a set of discrete actions. For example, a mod check may send a actionable modification code to one or more modifiers in response to a button press.

3. Modifiers

• Modifiers can modify anything about the player, player parameters, or the playback/recording data.
• Modifiers typically wait for a specific modification code before taking an action. For instance, the up-and-coming playback signal may be scaled in response to experimenter pressing a "louder" button during the mod check stage.
• Other modifiers may provide their own conditional checks and determine actionable events in lieu of a modification check. In these cases, there is little difference between a modification check and a modifier except perhaps when and how it is executed within the player

4. Analyses

• Analyses accept the returned data structure(s) from a player and analyze the data in some way. The nature of the analysis will depend on the experimenter's desired metrics.

In addition to these, SIN comes packaged with many higher-level functions for test control, subject creation, data handling, etc.

Why mod checks and mofifiers?

The decision to use a two-step modification procedure was made to allow users to mix and match test behaviors more easily. For example, the HINT scoring GUI and accompanying modification check were recycled with the MLST. The only difference between the two tests is the player parameters and the modifiers employed.

Installation and Software Dependencies

Step-by-step installation instructions of all core SIN software and dependencies are available at the link below. If you have any difficulty with insallation, please contact Chris Bishop at [email protected].

http://www.evernote.com/l/AWGfHzxf2PNGJ6j1YIb7q57vsgzQtCPZ56c/

Required Hardware

SIN can be run on any system running MATLAB using standard auditory equipment already present in many psychacoustic and clinical testing environments.

Required Hardware

1. A sound card with stereo in/out. ASIO drivers are a must.
2. Minimum of two stereo outputs.
3. Analog amplifiers (e.g., speaker or headphone amplifiers). These don't need to be terribly beefy, but DO need to have more power than you'd expect. We used CROWN XLS1000 speaker amplifiers and Tannoy Di5c speakers.
4. A computer running MATLAB v 2013b with a minimum of 12 GB of RAM. SIN is terribly inefficient in places and stores all data at double precision (32 bit). Thus memory requirements can be very high for lengthy tests.

Recommended Hardware

1. Equalizer(s) to flatten the frequency response of the playback loop.
2. A way to measure the SPL of the audio drivers being used (e.g., a SLM).

Available Tests

Currently, there are 48 individual tests or test segments designed and vetted in SIN. However, all of these tests fall into the following major categories. For a detailed description of each, please see SIN_TestSetup.m.

1. Hearing in Noise Test (HINT)

• Several variants of the HINT are available.

2. Multimodal Lexical Sentence Test (MLST)

• Auditory and audiovisual testing conditions are available and handled automatically through command line calls to Windows Media Player. CWB attempted to use more universal media players (e.g., VLC), but the command line execution routines had significant issues for testing purposes. For instance, VLC launches a new viewer with each command line call rather than piping a new stream to an existing window.

3. Reading Span

• The Reading Span is available through SIN, but uses a 3rd party EXE. The source is not available for the EXE so this test is more or less a "black box" in SIN's eyes.

4. Word Span

• An auditory working memory task implemented completely in SIN.

5. Acceptable Noise Level (ANL)

• ANL is a clinical noise tolerance test that has been implemented fully in SIN. All amplification and adjustments are handled digitally using a simple user interface.

6. Hagerman-Style Phase-Inversion Recordings

• Hagerman style inversion recordings are implemented fully in SIN.
• Note that as written this test requires a four-speaker setup. But it can be easily modified to accomodate fewer speakers. If this is something you'd like to do, contact CWB and he will point you to the line of code that must be altered in SIN_TestSetup.

7. System Calibration

• Sound Card Test: the goal of this test is to ensure that the output of each sound card channel is identical to all others and to provide a qualitative test for channel-cross talk. To do this, a standardized tone is played from all channels first, then each channel in turn. The user can use an independent signal monitor (an oscilloscope or perhaps even another sound card) to ensure that the signals are identical. In my experience, cross-talk is difficult to detect using hardware, but can typically be detected manually be cranking up the output on channels that should be "quiet" and listening to the output. CWB has used this approach to qualitatively test for channel cross-talk in CROWN amplifiers and FocusRite products.

• Calibrate Speaker Levels: this routine plays a sound file from each speaker in turn. The user can use calibration equipment and other hardware to ensure that this "standardized" sound is at the desired level. This absolute SPL then becaomse "0 dB" in SIN. All other stimuli should be calibrated relative to this calibration file. As written, SIN plays a speech-shaped noise sample used with the HINT.

• Audio Test (10 Hz click train): tests the relative levels and timing precision of the recording loop. With symmetical microphones, the levels should be matched to within tolerance limits (1 dB in SIN, but this may be impractical in some testing environments). To ensure the timing of playback/recording loops is adequate, a 10 click train is presented 10 times. These individual sweeps are then realigned. If realignment is necessary (i.e., there's jitter in the system), a warning is provided. Tolerance limits for timing precision should generally be very low (0 sample jitter in SIN). This may not be possible with all hardware/software setups.

• Weight Estimation: used to estimate the relative power levels between each speaker (or other audio driver) and recording channel. If two microphones are positioned in a listener's ear canals, then this is approximately equivalent to a binaural room impulse response (BRIR) collapsed across time and frequency. This procedure proved useful in Hagerman style signal extraction routines to adjust the theoretical SNR to account for head shadowing effects in multi-speaker arrays.

Developing New Tests

To illustrate a typical development work flow for a new test, consider the (abbreviated) development of the Hearing in Noise Test (HINT)

• Create a GUI that can be used to score each sentence of the HINT (see HINT_GUI)
• Create a mod check that calls the GUI and uses its output to decide if any actionable event is required (e.g., increase the SNR on the subsequent trial). See modcheck_HINT_GUI
• Create a modifier to implement the designed modification. In this case, modifier_dBscale_mixer changes the mixing matrix to alter the SNR in the next trial.
• Select a playback list. In this case, the playback list is one or more HINT lists.
• After all items in the playback list have been presented, analyze the data using an analysis script. See analysis_HINT.

Function List and Descriptions

Function	Description/Notes
addnoise2HINT.m	Function to add noise to HINT stimuli. User must provide a wav file with the noise. Assumes HINT stimuli are in the ../playback/HINT directory (hard coded for ease).
algo_HINT1up1down.m	Function to generate 1 up 1 down algorithm behavior for Hearing in Noise Test (HINT). This is essentially a wrapper for algo_staircase, which makes the actual decisions. algo_HINT1up1down just massages the 'score' returned by modcheck_HINT_GUI into an appropriate format and creates a (hard-coded) decision matrix for use with algo_staircase.
algo_HINT3down1up.m	Similar to algo_HINT1up1down.m, but with three-down-one-up staircase.
algo_HINT4down1up.m	Similar to algo_HINT1up1down.m, but with four-down-one-up staircase.
algo_HINTnochange.m	Placeholder algorithm that simply returns a mod_code of 0.
algo_NALadaptive.m	Not used by SIN any longer. This should be written in OOP fashion rather than using persistent variables. Simultaneous and parallel tracking algorithms are not possible as written.
algo_staircase.m	his is the core control for all stair case algorithms (e.g., 1up1down, 4down1up, etc.) in the SIN package. In order to remain flexible, it uses a "decision matrix" (see below) to provide instructions on what should be done on the next trial. Although this function controls the key decision making portions of staircase algorithms, it does not invoke any changed directly. Rather, it provides a general set of instructions through the decision matrix that can be interpreted in other functions.
align_timeseries.m	Function to align time series using one of several realignment strategies. If the signal in Y occurs before the signal in X, then Y is delayed to match Y. If the signal in X occurs before the signal in Y, then X is delayed to match X. Both time series are zero padded after realignment (either prepended zeros or appended zeros depending on the direction of the shift) so signal lengths match after realignment.
amp2db.m	Convert amplitude to decibel scale.
analysis_ANL.m	Analysis function for ANL testing.
analysis_AudioTest.m	Analysis function for AudioTest.
analysis_Hagerman.m	Analysis function for Hagmeran. This is still being actively developed and tested, particularly with respect to attenuation estimation.
analysis_HINT.m	Analysis for HINT-like tests. Includes several scoring schemes (traditional, reversals, etc.)
analysis_MLST.m	Analysis for MLST.
analysis_weight_estimation.m	Analysis for weight estimation tests.
analysis_WordSpan.m	Analysis for word span.
ANL_GUI.m	A GUI used to administer the ANL. This is also used by several other tests that are not strictly ANL.
ANL_modcheck_keypress.m	Early version of modcheck for ANL. No longer used.
aw_rms.m	Calculates A-weighted RMS. This function is borrowed from elsewhere.
calcBandSNR_v2.m	Donated by James Lewis.
class2txt.m	convert various classes to a descriptive text string. Designed to be used in combination with Selection_GUI and SIN_select.
color2colormap.m	Simple lookup table to create a colormap from typical 'color' values from plot, etc. (e.g., 'r', 'b', 'k', ...). This proved useful when creating bar plots that follow the same color scheme as ERPLAB's plotting functions.
comp_struct.m	Compare two structures. Borrowed from MATLAB file exchange.
concat_and_match_spectra.m	No longer used since it implicitly calls SIN_matchspectra, which has been replaced by match_spectra.m
concat_audio_files.m	This function will load and concatenate all audio files. Parameters below apply some control over stimulus processing (e.g., trimming silence from beginning and end of sound, etc.).
concatenate_lists.m	This function converts a LIST x N cell array to a (LIST x N) x 1 cell array. This proved useful when converting the output from SIN_stiminfo to a single file list.
create_signaltag.m	Function to create a repetitive train of a given signal with a specified SOA. Originally written for use with Hagerman-style recordings.
create_wordspan_lookup.m	Creates an XLSX lookup table to be used with SIN.
createHagerman.m	Function to create stimuli for Hagerman recordings. The basic idea is to provide a method of estimating SNR, a flag for holding the noise or speech track constant, and a naming scheme to use to save the output files. Hagerman stimuli for Project AD: HA SNR are comprised of ~1 minute of concat_target_trackenated HINT sentences (e.g., List01 - List03) paired with one of several noise types (speech-shaped noise (SSN) and ISTS). Other flags provide information regarding the number of noise channels to create and what time delay should be applied to each of the noise channels - recall that SSN and ISTS are single channel files.
createHINTlookup.m	Very basic function to create a lookup table for HINT. This basically uses Wu's originaly HINT lookup table (well, a manually edited version of it anyway ... CWB removed some of the file name information), and appends a suffix to the file names. This allows modcheck_HINT_GUI to successfully lookup sentence information.
createMLSTlookup.m	This function creates a suitable lookup table for MLST stimuli. This uses the provided MLST word list (sent to CWB by Wu) to create something closer to the HINT lookup table that is more robust.
db2amp.m	Convert decibels to amplitude (magnitude)
erplab_linespec.m	Function to kick back default color and line style values similar to ERPLAB. Not tested to see whether or not colors actually match precisely, so take color with a grain of salt. Code cannibalized from ploterps.m (part of ERPLAB).
fade.m	Function to fade sounds in/out by applying a window over a specified time range. Uses MATLAB's "window" function to create the fade in/fade out ramps.
fftplot.m	Plot amplitude spectrum of time series.
fillPlaybackMixer.m	Function to create a mixer of appropriate size for the playback device. This proved useful when using SIN on machines with different devices (e.g., 8-channel playback using the Halo Claro and two-channel sound playback using standard two channel cards).
filter_table.m	This function returns a filtered version of the original table that satisfies all user-specified criteria. The criteria are essentially fields and their corresponding acceptable values. Note: this is a more generalized version of import_HINT.m
filterA.m	Not used by SIN.
findcell.m	Borrowed from ERPLAB's deprecated functions. Thanks to Javier!
getCases.m	Borrowed from MATLAB file exchange.
getMatchingStruct.m	Function to find structures that match fieldnames and values provided by the user. This proved useful when trying to identify a specific modifier within a group of modifiers, but can be used to match any field within any set of structures.
git_is_current.m	Function determines if a repository is fully up to date with all changes committed or not.
git_sha.m	This function returns the SHA key for a GIT repository using the git.exe. Note that git.exe must be in the system path for this to work properly.
Hagerman_getsignal.m	Function to extract a signal from a pair of (phase inverted) Hagerman recordings. The non-phase inverted signal is returned along with its samppling rate. Note that the function assumes that the data are already time aligned. If this is not the case, then call align_timeseries.m before calling Hagerman_getsignal.m. If the signals are not temporally aligned, the output will be meaningless.
Hagerman_record.m	One of the first files developed for SIN. This was intended to be a standalone script to perform Hagerman-like recordings. CWB later modularized these routines to play nicely with player_main.m This is no longer used.
HINT_chooseAlgo.m	Function to select the current algorithm being used for spech in noise testing (e.g., HINT, MLST) on a given trial.
HINT_GUI.m	HINT scoring GUI. Recycled for MLST.
HINT_score2binary.m	Function to convert scoring arrays to a binary outcome for use with HINT and other similar tests. This is designed to work specifically with the scoring matrix generated by modcheck_HINT_GUI, but should be adaptable to other circumstances without too much trouble.
importHINT.m	Function to import HINT information from XLSX file. Can also return information specific satisfying search criteria (under development). If not search criteria are provided, then information for the entire list is returned. Note: CWB recommends using load_lookup_table.m instead.
Instructions.m	Code for Instructions GUI.
is_reversal.m	This function takes a time series, data, and determines which points are likely reversals (e.g., a reversal in the direction of change). This was written and intended to be used in combination with tests like the HINT (SNR-80 ...) and other adaptive algorithms that are required to terminate after a specific number of reversals.
iseegstruct.m	Borrowed from ERPLAB
iserpstruct.m	Borrowed from ERPLAB
label_datapoint.m	Adds a text label to a point in a figure. Does some basic checks to make sure the text is generally visible, but uses some plausible hard-coded values to do so. Might need to be more dynamic.
lineplot2d.m	Function to create two-dimensional line plots complete with error bars (if specified). CWB found that he was recycling a lot of code to create similar line plots in various functions for project AA, so decided to centralize the plotting routines.
load_lookup_table.m	Load a lookup table and return a table structure. This is how CWB should have written importHINT, but he didn't have the time to do it properly. SO, now he gets to rewrite it and deal.
make_stimuli.m	Script to make calibrated stimuli for all (current) SIN Tests.
map_channels.m	This function creates a channel mapping matrix used by PsychPortAudio to do appropriate sound mappings. Use iterative calls for playback/recording sound setup since these are currently handled in different mat files.
match_spectra.m	This function can be used to match the spectra of two time series. Below is an approximate outline of the procedure implemented here. Please note that there are many ways to do spectral matching and this may not be the best solution for your specific circumstance. This function uses a spectral estimator (pwelch) to estimate the PSDs of two time series and then attempts to filter these time series such that the PSDs are a close approximation of each other. Note that SIN_matchspectra provides an alternative method, but tends to be very slow and generally oerfits the data. CWB wrote this as a faster, more streamlined, and more principled alternative.
mex_uuid.m	Borrowed from MATLAB file exchange.
mixer2pan.m	Function to convert a DxP mixer to a pan string used by the pan parameter in FFmpeg. No longer used.
MLST_makemono.m	Function to set one of the audio channels in MLST MP4 files to 0. MP4s associated with MLST are stereo by default. We only want the audio to play speech out of a single speaker. To do this, we'll edit the MP4s using FFmpeg. No longer used.
modcheck_ANLGUI.m	Modcheck function coupled with ANL_GUI. ANL_GUI is a simple "run", "pause", "stop" GUI. This function returns codes compatible with modifier_PlayControl based on toggled buttons
modcheck_HINT_GIU.m	Modcheck for HINT.
modcheck_wordspan_gui.m	Modcheck for Word Span.
modifier_append2playlist.m	Appends new playback data to the playback list in player_main. This proved useful when using adaptive algorithms (e.g., 3down1up for SNR-80 estimation) that may require more stimuli than originally specified by the user. This essentially just calls SIN_getPlaylist with the user-provided settings as written the SIN options structure. These values are then appended to the 'stim' variable in player_main. Note: This function intentionally ignores values in the "files" field of specific.genPlaylist. The assumption is that we're looking for a new playlist to append to the playback list in player_main
modifier_dBscale_mixer.m	Function is designed to apply a decibel scaling factor to the invoking player's mod_mixer. It's designed to be used with the portaudio_adaptiveplay player. The function ultimately modifies the d.player.mod_mixer field.
modifier_exit_after.m	This is sort of a wrapper used to check multiple exit criteria (modifiers). For instance, if the user wants to exit only after 7 reversals and a minimum of 20 trials, then this is the function to use.
modifier_exit_after_nreversals.m	This function will set the player state to 'exit' after a specific number of reversals are encountered in a specific data/physical channel combination. This must be paired with modifier_trackMixer in order to work since it relies on the subfield created by _trackMixer to compute the number of reversals.
modifier_exit_trials.m	This will set the player state to exit based on the number of trials. Several possible parameters available, including : trial minimum, trial maximum.
modifier_exitAfter.m	Function that sets player status to "exit" after an arbitrary expression evaluates to "true". This is designed to work with "player_main.m". Note: users should use modifier_exit_after.m instead.
modifier_instructions_by_trial.m	Presents instructions using the "Instructions" GUI on a specific trial.
modifier_NALscale_mixer.m	Function to query and apply the appropriate decibel (dB) scaling factor from the (currently running) algo_NALadaptive function to the appropriate elements of the player's mod_mixer. No longer used.
modifier_PlaybackControl.m	Function designed to handle basic playback control requests. At time of writing, this includes "pause", "run/resume", and "exit/quit".
modifier_ShowInstructions.m	Wrapper to present basic instructions during player startup. Designed to be paired with Instructions.fig. This specific "modifier" is a bit of a misnomer. It doesn't actually modify anything. It just presents instructions at the beginning of each test.
modifier_trackMixer.m	Function to track mod_mixer settings. This is done by appending the current state of the mod_mixer to a variable established in d.sandbox.
player_main.m	This is the main work horse of SIN. It coordinates all playback/recording/modchecks/modifiers. It's the GLUE!
plot_psd.m	Plot the power spectral density (PSD) of a time series or multiple time series. Spectral estimation computed using Welch's periodogram (see doc PWELCH for details).
plot_waveform.m	Simple function to plot a time varying waveform in MATLAB.
portaudio_GetDevice.m	Function to find and return device based on input information. Input information can http://docs.psychtoolbox.org/GetDevices. Function also provides a basic device check to make sure the device information is not out of date.
recordings_to_file.m	This function accepts a results structure from player_main and writes the recordings to file with the write parameters provided by the user. Recordings are written to file in the .wav format at the sampling rate used for the recordings during stimulus testing.
regexpdir.m	Borrowed from matlab file exchange.
remixaudio.m	This function is designed to remix audio (e.g., create arbitrary linear combinations of channels in an existing file). Also supports other features during remixing, including introducing arbitrary temporal shifts in the data (using circshift). The latter was especially useful when converting 1 noise channel into multiple, uncorrelated channels for Hagerman recordings.
runSIN.m	Main SIN GUI. Run this at the command line to get started.
Selection_GUI.m	A simple selection GUI. Used in lots of contexts.
sem	Compute standard error of the mean of X. CWB performs this computation frequently and decided to write a function to do it (finally) to save the copy and pasting job he's been doing for half a decade. If DIM is not defined, sem will operate on the first non-singleton dimension. Note that STD is estimated using an unbiased estimator (flag=0). See doc STD for more information.
SIN_assignUUID.m	SIN uses a Universally Unique IDentifier (UUID) to track individual tests and test sequences. This function assigns UUIDs in a semi-intelligent way by ensuring that individual tests receive a UUID and all tests within a test sequence share a common UUID. This should allow the user some unimbiguous information to quickly pool results.Uses uuidgen.m to generate the UUID. uuidgen can be downloaded here: http://www.mathworks.com/matlabcentral/fileexchange/21709-uuid-generation/content/uuidgen.m
SIN_CalAudio.m	The calibration procedure for SIN sets the HINT-Noise.wav (a speech shaped noise sample) to 0 dB and scales all other stimuli or stimulus sets to also rest at 0 dB. The user can then present the HINT-Noise stimulus through their sound playback system and adjust (via hardware) the sound pressure level (SPL) to the desired level. Following this procedure, all stimuli/stimulus sets should be have a nearly identical SPL, provided that the frequency response of the playback/recording loop is flat (enough). CWB recommends using hardware (e.g., a graphical equalizer) to flatten the frequency response of your playback/recording loop.
SIN_ClearErrors.m	Function to assist in clearing and recovering from errors. This will likely grow as SIN does and more and more error types are encountered. For now, though, SIN will simply clear PsychPortAudio, close figures,and close open 'Screens'
sin_gen.m	Generate a sine wave. Borrowed from Jess Kerlin.
SIN_getPlaylist.m	Function to return a playlist for a specific test. This is essentially a wrapper for SIN_stiminfo, which returns directory and file information.
SIN_getsubjects.m	Query existing subjects. Query based on existing directories. If the directories don't exist, then we have to assume that no subject data has been collected. Code modified from: http://stackoverflow.com/questions/8748976/list-the-subfolders-in-a-folder-matlab-only-subfolders-not-files
SIN_gettests.m	Function to return tests for a given subject.
SIN_keywords.m	In SIN, key words are capitalized and word options are enclosed in square brackets and separated by a forward spash (e.g., [are/were] or [ARE/WERE]). Words are separated by white space.
SIN_load_results.m	This function loads SIN results structures listed in file_names.
SIN_loaddata.m	Generalized function to load data from common file types used in project AA (and probably others) as well as generate standardized data format for data matrices, provided some sensible assumptions are met. CWB wanted a function that would accept virtually any commonly used input data type (file names, data structures, etc.) of known format and massage the data into a common format that can be used by other functions for analysis and plotting.
SIN_makeFilter.m	This function is no longer used by anything in SIN.
SIN_maskdomain.m	Function returns a logical mask of the domain within the specified bounds. This is something CWB had to do repeatedly when computing SNR in frequency and time domains. CWB preferred to have one central function do the masking so the masking is always done the same way and it's easy to introduce the operation into new functions.
SIN_matchspectra.m	Spectrally match two time-series in the frequency domain. The algorithm is horribly inefficient and prohibitively slow with even modest time series. I recommend using match_spectra.m instead.
SIN_recoverDevice.m	This function attempts to recover a sound playback/recording device in the event that its DeviceIndex field has changed. This occurs frequently and most notably when sound hardware is added or removed (e.g., a USB device) or (I think) when users change their sound playback/recording settings through Windoze.
SIN_register_subject.m	Function that will (hopefully) handle subject registration cleanly. This should be rewritten as OOP.
SIN_removepunctuation.m	Function to remove punctuaion from a string. Currently removes the following characters
SIN_review_recordings.m	This routine uses a GUI to select and plot out recordings acquired saved in a results structure from player_main.
SIN_runAnalysis.m	Function to run SIN analyses using the analysis subfield of the SIN options structure.
SIN_runsyscmd.m	Wrapper function to handle and run system commands. This is used as a "player" to run executables and the like.
SIN_runTest.m	Master control function to run various tests associated with SIN. The basic idea is to pass the function a unique test identifier (CWB is thinking a string) that can be used to execute a specific set of instructions. The upshot of this approach is that the same test can be executed with ease from the commandline or via a GUI.
SIN_saveRestuls.m	Basic function to save results. This was originally written to save results from portaudio_adaptiveplay, but shouldn't be difficult to expand to various results formats.
SIN_select.m	This creates a simple GUI to aid the selection of various types of options. Originally designed as a helper function for the dynamic selection of playback and recording devices through PsychToolBox.
SIN_stiminfo.m	Function to return stimulus lists (directories, filenames) for the tests used in the SIN suite.
SIN_TestSetup.m	Function to return test information. This will vary based on the test. Alternatively, this can also return a list of available tests (default).
SIN_UsedListInfo.m	Function to track and return information regarding the used list structure. Most methods below load, modify, or save a structure containing information on which lists have been used and with which tests. This structure is described here as the "UserList structure". This is, however, a n N x 2 cell array, where N is the number of lists that have been used. The first column contains the directory name of the list. This is how lists are "tagged". Each list must be in its own directory. If the directories are MOVED and previous calls use an absolute path, then the function will think the same stimuli are different lists. Silly, but it's the most robust way CWB could come up with in a hurry. Note: Should be rewritten as OOP.
sort_results_by_time.m	Sort SIN results structures by time. This function proved useful when trying to find the most recent preceding file relative to a moment in time.
struct2keyval.m	Converts a structure into key/value inputs. This is essentially the counter part to varargin2struct. Function is still a bit basic, but should be able to expand on this to work with many data types.
threshclipaudio.m	Function designed to remove silent periods from beginning and end of a sound file. To do this, the code removes all samples preceding the first sample that exceeds the provided (absolute) amplitude threshold. The end is clipped such that all samples following the last sample that meets or exceeds the provided amplitude threshold are discarded.
timestamps.m	Creates time stamps for a time series of length N.
update_results.m	Function loads all results for all subjects and rewrites them by calling SIN_saveResults. This in effect updates the saved results structures to match whatever the current format is. This has proved useful when adding additional variables to save to the results structure (e.g., end_time).
uuidgen.m	Borrowed from MATLAB file exchange.
uuidgen_cimp.cpp	Borrowed from MATLAB file exchange.
uuidgen_cimp.mexa64	Borrowed from MATLAB file exchange.
uuidgen_cimp.mexglx	Borrowed from MATLAB file exchange.
uuidgen_cimp.mexw32	Borrowed from MATLAB file exchange.
varargin2struct.m	This function massages input parameter pairs (e.g., 'dbStep', [1 23]) into a structure with the appropriate field names. Function will also handle various input types, such as pairing structures and input parameters. This proved useful for the Speech In Noise (SIN) suite CWB wrote.
wordspan_find_keywords.m	This will load the provided word span stimulus files and return just the time trace of the keywords within each sentence. This proved useful when trying to calibrate the word span since the carrier phrase does not vary (much) from one sentence to the next. We will need to build in some sanity checks on the carrier phrase from sentence to sentence to make sure it hasn't changed fundamentally in some way (e.g., through error or other machinations).
wordspan_rename_files.m	Renames all files used for word span.
WordSpan_Scoring.m	Scoring GUI for Word Span