Public Safety Audio Quality
Modified Rhyme Test
PURPOSE
The PSCR (Public Safety Communications Research) program is working with fire fighters to understand how background noise affects voice communications. The goal is to learn what technology and operational improvements may help overcome background noise issues.
METHOD
For the testing described here, the PSCR Audio Lab used a subjective intelligibility test called the MRT. The test measures correct responses, rather than subjective judgment (i.e., is audio intelligibility bad, poor, good, or excellent?) as in a MOS. Instead, in the MRT, listeners were asked to select from a list the word they believe they heard from the talker.
The listener was given a PDA. The listener played a randomly served sentence. After each sentence the PDA displayed a request to choose the word heard. For example:
“Please select the word _______.”
bed | led | fed |
red | wed | shed |
Fifty groups of six words were spoken by three female and three male talkers. Thirty listeners participated in the test, each of whom heard five lists of 300 sentences.
NOISE CONDITIONS
The MRT comprised nine noise conditions — either with or without an SCBA mask with no additional noise or noise from a fire truck pump panel, a rotary saw cutting metal, a chainsaw cutting wood, a PASS alarm, or a low-air alarm.
For complete details about how the test was run, see NTIA Technical Report TR-08-453.
Click here to listen to a few samples from the testing.