OPM-MEG Audio Latency Measurement

Tool to measure and characterize the trigger-to-ear audio latency of an OPM-MEG stimulus pipeline using a BBTK (Cambridge Research Systems Black Box ToolKit v3). Compares VPIXX hardware-scheduled audio to PsychoPy PTB software audio, across both pneumatic-tube earpieces and room loudspeakers.

Also includes a demo script for those that want to use VPIXX to play audio.

Contents

• audio_latency_bbtk.py — PsychoPy script that plays 200 trials of a short stimulus (sine, click, or narrow-band noise) with a simultaneous pixel-mode trigger to BBTK. Supports two audio paths: vpixx (DATAPixx3 audio schedule, vsync-locked) or psychopy (PTB backend, prescheduled to next vsync).
• analyze_click_latency.py — analyzer for four BBTK RTL files in a 2×2 design (audio path × transmission medium). Produces summary tables in the console, a full 2×2 analysis figure, a tail-filtered variant, and a sharable single-page directives table for researchers.

Software Requirements

• PsychoPy 2026.1+ (for the measurement script)
• pypixxlib (VPIXX)
• numpy, matplotlib (for the analyzer)
• The BBTK Software Suite for recording and exporting RTL files

Hardware Requirements

• A stimulus control PC, DATAPixx3, ProPixx projector (120Hz).
• Audio delivery system such as speakers or SOUNDPixx pneumatic tubes.
• DATAPixx3 DOUT wired to a BBTK TTL input to receive PixelMode triggers.

Setup

VPIXX DOUT (Pixel Mode B) --> BBTK digital input 1  (pixel_trigger)
Microphone at earpiece    --> BBTK mic channel 1

Audio source depends on AUDIO_PATH:

• AUDIO_PATH = 'vpixx' → VPIXX audio-out → SOUNDPixx amp → tubes / speakers
• AUDIO_PATH = 'psychopy' → computer → DAC → amplifier → tubes / speakers

Running a measurement session

python audio_latency_bbtk.py

The startup dialog prompts for measurement_id, audio_path (vpixx or psychopy), and tone_type (sine, click, or noise). Recommended default: 2 kHz click, which has the sharpest onset for BBTK detection.

Each session records 200 trials at 1.8–2.2 s jittered ISI. Duration is about 7 minutes.

Export each BBTK recording as an RTL file.

Analyzing

Edit the INPUT_FILES dict at the top of analyze_click_latency.py to point to your four RTL files. Set SPEAKER_TO_MIC_DISTANCE_M to your measured chair-to-speaker distance (used to correct for open-air transit in the speakers baseline). Set FRAME_MS to your display's frame period (8.33 ms for 120 Hz) — this is used to detect the one-frame VPIXX race tail. Then run:

python analyze_click_latency.py

Notes

VPIXX audio playback via DPxWriteRegCacheAfterVideoSync has a race condition with win.flip() that causes ~10–15% of trials to fire one display frame later than the main mode. The analyzer detects this bimodality and reports both the raw SD (including the tail) and the main-mode SD (excluding it). Use the main-mode SD as the underlying hardware precision; the tail is a software artifact separable from audio-path physics.