Run Audio Experiment

runAudioExperiment is an utility to run an experimental setup where input times must be recorded with precision synchronized to an audio stimulus. The utility was developed in the context of "Simple and cheap setup for measuring timed responses to auditory stimuli" (Miguel et. al., under review).

The utility works by executing several trials that consist of reproducing an auditory stimulus while simultaneously recording the signal of an input device. As an output, the utility saves an audio file for each trial. With the expected setup, each audio file has the input signal in one channel and the stimulus signal in another. Recording both signals (the input and the stimulus loopback) allows extracting the input times relative to the stimulus signal by synchronizing the recording with the orignal stimulus (see utility rec2taps).

For details on how to create the setup, please see Miguel et. al., under review.

Installation

• From pypi:

    pip install m2-runAudioExperiment

• From sources:

    git clone https://github.com/m2march/runAudioExperiment.git
    cd runAudioExperiment
    python setup.py install

Once installation is done, the script runAudioExperiment should be available in the command line interface.

Usage

runAudioExperiment trial_config stimuli_list output_dir

• trial_config is a path to a yaml format file describing the configuration of the execution
• stimuli_list is a path to a txt file declaring the path to the stimuli audios, one per line, relative to the current dir
• output_dir is a path to a directory where output from the experiment will be written. In case the directory does not exists, it will be created. If the directory exists, it must be empty.

Trials

Each trial execution consists of the following steps:

• A black screen is shown for a specified duration
• Screen changes color and white noise with a tone is played
• A black screen is shown and the stimulus is played while recording the input device. A silence time after the stimulus ended can be configured.
• Screen changes color and white noise with a tone is played

Trials are executed back to back.

Configuration

The following is an example of a trial configuration file:

black_duration: 600         # Duration of black screen (in ms)
c1_duration: [1500, 3000]   # Duration of first noise screen (in ms)
c1_color: "#afd444"         # Color of first noise screen
c2_duration: 000            # Duration of second noise screen (in ms)
c2_color: "#afd444"         # Color of second noise screen
randomize: false            # Whether trial order should be randomized
sound_device: "USB Audio"   # String or int identifying the sound deviced used
silence_duration: 1500      # Duration of silence after stimuli playback

Durations may be specified as a single number or as a list of two numbers. If it is a single number, it specifies the duration of the section. Otherwise, the duration will be a random number between the two defined ones. If the duration is 0, the section is skipped.

sound_device is a string used to locate the sound device to be used. In our proposed setup, an external USB card is usted. A list of available devices can be obtained by executing:

`runAudioExperiment -l`

Output

After the experiment is executed, the following files are generated in output_dir:

• experiment_settings.json: a copy of the configuration defined in trial_config in json format. The configuration is extended with the complete information of the selected sound device.
• trial_settings.csv: a csv formatted table describing the executed trials. This table allows knowing the duration of the sections where duration was specified as an interval. It also presents the order in which the stimulus were presented in case the randomize setting was set to true.
• *.rec.*: a serie of audio files are generated, one per trial, with the recording of the input signal and the stimulus lookback. The files are prefixed with the basename of the stimulus file and end with a .rec extension continued with the original extension of the stimulus file.

Example execution

The directory example contains example data to test run the utility and see its workings and results. A test run can be done as so:

$> cd example
$> runAudioExperiment test_config.yaml stimuli_3.txt output

This will run the experiment with three stimuli and create a new folder named output were recordings and trial_settings.csv will be output.

Before using the utility it may be necessary to configure the sound_device configuration in test_config.yaml to refer to a sound card available in the computer. Use runAudioExperiment -l to list options.

Caveat in the duration of the sections

The duration of the sections may differ slightly from the specifications, mainly due to the delay in starting playback in the soundcard. Added delay is around 20ms. The --debug-durations flag instructs the utility to write ohe actual duration of the sections in an extra trial_durtions.csv file.