The study investigated the effects of noise, one of the major environmental stressors, on speech perception and spoken word comprehension. Three tasks were employed – listening span, listening comprehension, and shadowing – in order to find out to what extent different types of background noise affected speech perception and encoding into working verbal memory, as well as spoken word comprehension. Six types of maskers were used – (1) single babble masker in English, (2) single babble masker in Mandarin, (3) multi babble masker in Greek and (4) construction site noise, (5) narrow-band speech signal emulating phone effect and (6) reverberated speech signal. These could be categorized as energetic (2, 3, and 4), informational (1) and signal degradation (6 and 7) noise maskers. The study found that general speech perception and specific word comprehension are not equally affected by the different noise maskers – if shadowing is considered primarily a task relying on speech perception, with the other two tasks considered to rely on working memory, word comprehension and semantic inference. The results indicate that informational masking is most detrimental to speech perception, while energetic masking and sound degradation are most detrimental to spoken word comprehension. The results imply that masking categories must be used with caution, since not all maskers belonging to one category had the same effect on performance. Finally, introducing a noise component to any memory task, particularly to speech perception and spoken word recognition tasks, adds another cognitively stimulating real-life dimension to them. This could be beneficial to students training to become interpreters helping them to get accustomed to working in a noisy environment, an inevitable part of this profession. A final study explored the effects of noise on automatic speech recognition. The same types of noise as in the human studies were tested on two automatic speech recognition programs: Otter and Ava. This technology was originally developed as an aid for the deaf and hard of hearing. However, their application has since been extended to a broad range of fields, including education, healthcare and finance. The analysis of the transcripts created by the two programs found speech to text technology to be fairly resilient to the degradation of the speech signal, while mechanical background noise still presented a serious challenge to this technology. / Dissertation / Doctor of Philosophy (PhD) / The study investigated the effects of noise, one of the major environmental stressors, on speech perception and spoken word comprehension. Throughout three different tasks (listening span task, in which participants were asked to recall a certain number of items from a list; listening comprehension task, in which listeners needed to demonstrate understanding of the incoming speech; and shadowing, in which listeners were required to listen and simultaneously repeat aloud the incoming speech), various types of background noise were presented in order to find out which ones would cause more disruptions to the two cognitive processes. The study found that general speech perception and specific word comprehension are not equally affected by the different noise maskers – provided that shadowing is considered primarily a task relying on speech perception, with the other two tasks considered to rely on working memory, word comprehension and semantic inference, or the way the listener combines and synthesizes information from different parts of a text (or speech) in order to establish its meaning. The results indicate that understandable background speech is most detrimental to speech perception, while any type of noise, if loud enough, as well as degraded target speech signal are most detrimental to spoken word comprehension. Finally, introducing a noise component to these tasks, adds another cognitively stimulating real-life dimension, which could potentially be beneficial to students of interpreting by getting them accustomed to working in a noisy environment, an inevitable part of this profession. Another field of application is the optimization of speech recognition software. In the last study, the same types of noise as used in the first studies were tested on two automatic speech recognition programs. This technology was originally developed as an aid for the deaf and hard of hearing. However, its application has since been extended to a broad range of fields including education, healthcare and finance. The analysis of the transcripts created by the two programs found speech to text technology to be fairly resilient to a degraded speech signal, while mechanical background noise still presented a serious challenge to this technology.
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/27985 |
| Date | January 2022 |
| Creators | Eranović, Jovan |
| Contributors | Stroińska, Magda, Cognitive Science of Language |
| Source Sets | McMaster University |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0019 seconds