Phonetic Analysis of Speech Signals Simultaneously Recorded at Ear and Tips
Herbert J. Oyer · The Journal of the Acoustical Society of America · 1958
Previous study has shown that when speech signals were recorded at lips and left ears of speakers, and fed to headsets of trained listeners at specified S/N ratios, signals, origin ear, were significantly more intelligible than signals, origin lips. A comparative analysis is made of phonetic elements of listener response to speech signals, origins ear and lips elicited in the earlier study. Listener responses totaled 72 000. Listeners highly trained on fifty words tend to restrict substitute responses to words within the list for signals of both origins. Fewer substitutions of words from outside the original list occur as noise in the listening condition becomes more destructive. Substitutions remain relatively constant between origins through ratios. Omissions for stimuli, origin lips, increase more rapidly. Number of sounds in the test stimuli had no differential effect on intelligibility when related to origin. The frequency of occurrence of two, three, four, and five sound word substitutes is in almost complete agreement with frequency of occurrence of two, three, four, and five sound words in the test list. Analysis of intelligibility errors relative to vowels and diphthongs in the test list revealed that words containing (æ, ∧, α, ɔ, aʋ, and ɔl) were less intelligible from origin ear. than origin lips through S/N ratios. [This research was carried out in connection with Contract AF 19(604)-1577, monitored by Operational Applications Laboratory, Air Research and Development Command, Bolling Air Force Base, Washington 25, D. C.]