The Effects of Degraded Audio on Memory for Speech Passages

Abstract

As the US migrates to digital radio, broadcasters are making important decisions about the transmission rates they use on primary and secondary audio channels. One way to evaluate coders at different bit rates is to elicit consumers' opinions of the audio quality. However, it is evident from several lines of research that consumers' preferences do not always match their ability to process information efficiency. Therefore when evaluating coders at low bit-rates it is important to measure consumer's processing efficiency as well as examining their preferences. With regard to audio, past research has shown that degraded speech, including speech masked with noise and synthetic speech, impairs individuals' ability to recall content. The current research explores whether speech coded at very low bit rates (i.e., 9 kbps and 24 kbps) also impairs consumers' memory for text passages. Concurrently, we examine people's subjective ratings to see whether these ratings correlated with our objective recall measure. Results suggest that participants' recall of information dropped significantly at 9 kbps, a finding which has important implications for broadcasters. Participants' audio quality ratings did not correlate with their ability to remember story details, supporting the notion that merely asking consumers to rate the quality of a signal may not provide a complete picture of how impaired audio will affect their behavior. However, interestingly, their professed level of interest, comprehension, and enjoyment with the story did correlate with their memory scores, making these questions better candidates for subjective tests of very low bit rate coders.