>Psychtoolbox>PsychTests

KeyboardLatencyTest([triggerlevel=0.01][,modality=0][,submode][,portString][,audioDevice])

Uses sound capture with high timing precision via
PsychPortAudio() for measuring latency, timing accuracy and variability
of various response input devices, e.g., keyboard, keypad, mouse, other
HID devices like Joysticks etc., various response button boxes, and some
exotic response devices.

Whenever the script tells you, hit a key on the tested input device
hard and loud enough so the noise of hitting the button or key can be
recorded by the attached microphone. This noise will be timestamped by
the code as the “true” button press time. Timestamps acquired
by standard input device queries are compared against that reference
and the difference is computed as device latency.

Sound is captured from the default recording device, waiting
until the amplitude exceeds some ‘triggerlevel’.

The ‘modality’ flag chooses between keyboard (==0 - the default), and
mouse (==1). ‘portString’ allows to select which keyboard to test on some
systems (OS/X and Linux). It also allows to select which mouse to test on
Linux, but not on other systems.

A ‘modality’ of 2 queries the keyboard, a keypad, a mouse, or other HID
devices by use of the KbTriggerWait() function. ‘submode’ specifies the
KbName() keyCode of the key to test on a keyboard. By default the SPACE
key is used. For other devices you *must* specify a key or button number in
‘submode’. Specifiying numbers of non-existent keys or buttons will cause
an infinite hang of Matlab or Octave, as there is no way to interrupt the
function or press a non-existent key or button. The optional ‘portString’
specifies the deviceIndex of the device to test. If omitted, the primary
keyboard is tested.

A ‘modality’ of 3 will test the PsychRTBox() driver for the USTCRTBOX reaction
time button box, if such a box is attached. A setting 4 will also test
that box, but without opening a connection to it, ie., it is assumed that
the box is already open from a previous call of this function with
modality setting ‘3’. This allows to repeat the test many times without
recalibrating the box. The optional ‘submode’ flag selects different ways
of testing the box: A setting of zero will perform a ‘ClockRatio’
calibration to provide exact live timestamps and to test drift
correction. A setting of 1 will skip this, so live timestamps will
exhibit clock-drift and only the post-hoc timestamps will be somewhat
exact. A setting of 2 will skip collection of post-hoc timestamps.

A ‘modality’ of 5 will exercise the RB-x30 response pads from Cedrus.

A ‘modality’ of 6 will exercise the PST serial response button box.
Setting ‘submode’ to 1 will optimize for use with FTDI serial-USB
converters.

A ‘modality’ of 7 will exercise the CMU serial response button box.
Setting ‘submode’ to 1 will optimize for use with FTDI serial-USB
converters.

A ‘modality’ of 8 will exercise the Bitwhacker emulated response button box.
Setting ‘submode’ to 1 will optimize for use with FTDI serial-USB
converters.

A ‘modality’ of 9 will exercise the fORP serial response button box in
program mode 0.
Setting ‘submode’ to 1 will optimize for use with FTDI serial-USB
converters.

A ‘modality’ of 10 will measure timing of touch input devices like
touchscreens or touchpads. ‘portString’ allows to select a specific
touch input device as enumerated via GetTouchDeviceIndices(). By default
the first detected touchscreen or touchpad is used. Note that on some
operating systems you must hit the touchscreen or touchpad inside the
displayed onscreen window, as those systems will not register touches
outside the onscreen window.

The optional ‘portString’ argument can be set to define the serial port
to connect to for response devices that are connected via serial port.
By default, the proper serial port is auto-detected, but you can override
a wrong detection this way.

The optional ‘audioDevice’ allows to specify the device index for the
sound card to use for audio capture, in case multiple soundcards are
installed in the machine. By default, a sound device is auto-selected,
which is capable of sound capture.

Obviously this method of measuring carries quite a bit of uncertainty
in exact timing, but with a high quality microphone, proper tuning and
good sound hardware, it shouldn’t be off too much. At least you get a
rough feeling for the lags inherent to keyboards and mice.