Objective

Assess the latency of the speech recognition pipeline on the robot.

Associated Requirements Untitled (https://heathered-english-a1d.notion.site/7bf922f46af548eba06b9167e41e23ba)
Equipment Stretch RE1
Elements HRI - Speech Recognition
Personnel Lead:
Location AI Makerspace, Tepper School of Business

Procedure

  1. Turn the robot on and home it.

  2. Activate the speech recognition pipeline.

  3. Activate the test script that measures the time between completion of voice command input and transcription.

  4. Praveen (patient) will speak (in a normal voice) a random sentence (from https://randomwordgenerator.com/sentence.php).

  5. The recognized voice will be transcribed. The time duration for this is measured using the test script.

  6. Steps 3-5 are repeated 10 times. The mean time for transcription is then calculated.

  7. Kill the script and turn the robot off.

Validation

Although our performance requirements have a desired latency of within 5 seconds, we are only partially testing the robot's speech system as a transcription to control module is to be implemented as well, and thus we require the robot to transcribe test within 2.5 seconds of completing the command. We will test this at a later point of time.

Testing Results and Evidence

Date of Completion: February 27, 2023

Results