| Objective: This test aims to set up a noise-signal test platform in complexlistening environments and Sichuanhua noise-signal simulation test, basedon which normal hearing in four-channel cochlear implant of simulationprocessing by Sichuanhua speech materials (SAMs), which are suitable forSichuanhua-speaking impaired hearing (HI) patients (including hearingaids (HAs) and cochlear implant (CI) patients) will be developed.Method:1) Collecting noise: Four types of noise, namely construction sitenoise, traffic noise, industrial noise, community noise, was collected onspot by high fidelity microphones.2) Recording noise:2hosts and2hostesses from Chongqing TV Station Satellite Channel and CountryChannel were invited to record Mandarin speech test materials (-ratio) andSichuanhua acoustic speech materials (SAMs) respectively, and1host and1hostess form Guangdong TV Station Cantonese Channel are invited torecord Cantonese acoustic speech materials (CAMs).3) Noise-signal testplatform in complex listening environments was set up via Tiger SpeechMicrosoft, a large speech testing platform designed by House Ear Institute; 4)4-CH CI simulation processing: By reducing sound quality of speechsignals via ICAST software, the distorted speech recognized by CI wassimulated.5)50NHs were recruited and their speech recognitionperformance was examined respectively in quiet state and in four-channelcochlear implant of simulation processing (4-CH CI simulation processing)state.6) Standard SAMs were developed.7) Sichuan dialect-speaker groupwho were tested in the first experiment were examined for their speechdiscrimination scores for the second time under the same conditions as thefirst experiments’with an interval of30days between two experiments.8)Speech recognition performance of20HAs and11CIs were examined inquiet state and in state with signal to noise ratio (SNR) at+15dB and-5dBrespectively.9) A one-way analysis of variance, with the test list as atreatment factor showed no significant effect for the test list. The standarddeviation across ten test lists ranged from1.35%to5.23%for individualsubjects, with a mean of3.30%. Linear regression analysis was performedamong monosyllable lists, bisyllabic word lists and sentence intelligibilityand the retest-test reliability was examined.Results:1) Noise-signal test platform in complex listening environmentswas set up.2) More effective noise-reducing software was designed andreliable experiment data was offered for designing the speech processor forChinese speakers.3) The following is the speech discrimination scores of both Sichuanhua-speaker Group and Mandarin-speaker Group undernoise-signal enviroment by applying Mandarin Speech Perception TestMaterials (MSPTM)and Sichuanhua acoustic speech materials (SAMs)respectively.①Speech discrimination score of NH:a. Mandarin-speaker Group----MSPTM: speech discrimination scoresfor monosyllable lists, bisyllabic word lists and sentence intelligibility were56.045%、58.874%、75.569%respectively.b. Sichuan dialect-speaker Group----SAMs: speech discriminationscores for monosyllable lists, bisyllabic word lists and sentenceintelligibility were62.673%、65.903%、81.223%respectively.c. Sichuan dialect-speaker Group----MSPTM: speech discriminationscores for monosyllable lists, bisyllabic word lists and sentenceintelligibility were58.201%、68.430%、81.737%respectively.d. Speech discrimination scores for all NHs under quiet state was100%. Linear relationship existed among speech discrimination scores formonosyllable lists, bisyllabic word lists and sentence intelligibility.e. Comparison between Sichuan dialect-speaker Group----SAMs andMandarin-speaker Group----MSPTM: There was no significant differenceamong and within monosyllable lists, bisyllabic lists and sentenceintelligibility (P＞0.05). f. There was significant statistic difference between monosyllable listsand sentence intelligibility (P＜0.01）. There was no significant differencewithin bisyllabic lists intelligibility. Linear relationship existed amongspeech discrimination scores for monosyllable lists, bisyllabic lists andsentence intelligibility.②Retest—test reliability of Sichuan dialect-speaking NHs: speechdiscrimination scores for monosyllable, bisyllabic and sentenceintelligibility were63.4857%,63.3776%,82.8369%respectively.Comparing these results with the data of the first test, there was nosignificant difference in disyllabic words lists (P＜0.05）, which indicatedthat the speech test was of great stability.③Sichuan dialect-speaking HAs:a. HA-SAMs: tested at quiet state, with SNR+15dB state, and withSNR-5dB statespeech discrimination scores for monosyllable intelligibility:87.933%、91.417%、81.217%，speech discrimination scores for bisyllabic intelligibility:91.195%、90.988%、83.724%，speech discrimination scores for sentence intelligibility:96.844%、91.238%、79.844%.b. HA—MSPTM: tested at quiet state, with SNR+15dB state, and with SNR-5dB statespeech discrimination scores for monosyllable intelligibility:98.090%、96.000%、88.263%，speech discrimination scores for bisyllabic intelligibility:98.734%、96.952%、80.882%，speech discrimination scores for sentence intelligibility:99.578%、99.685%、89.259%.c.①Applying SAMs and MSPTMto HA subjects respectively have greatimpact on their speech recognition performance,and there is significantdifference in speech recognition performance (F=18.292，P=0.000，P﹤0.05).②When subjects are tested at quiet state, at state with SNR+15dB,and at state with SNR-5dB state respectively, there is significant differencein speech recognition performance (F=54.265，P=0.000，P﹤0.05).③There is significant difference in speech recognition performanceamong HAs with mild, moderate and severe hearing loss (F=23.705，P=0.000，P﹤0.05).④Whether using different speech audiometry word lists and the extentof hearing impairing of HA patients before their wearing hearing aids havegreat impact on speech recognition performance and there is significantdifference (F=0.555，P=0.696, P﹥0.05).. ⑤Adopting SAMs and MSPTMrespectively, using different speechaudiometry word lists and when subjects are tested at quiet state, at statewith SNR+15dB, and at state with SNR-5dB state respectively, have greatimpact on speech recognition performance and show significant difference(F=4.622，P=0.001, P﹤0.05).④a. CI-SAMs: tested at quiet state, with SNR+15dB state, and withSNR-5dB statespeech discrimination scores for monosyllable intelligibility:75.333%、87.467%、76.400%，speech discrimination scores for bisyllabic intelligibility:87.771%、87.485%、73.714%，speech discrimination scores for sentence intelligibility:70.333%、87.666%、62.952%.b. CI—MSPTM: tested at quiet state, with SNR+15dB state, and withSNR-5dB statespeech discrimination scores for monosyllable intelligibility:90.000%、85.600%、72.769%，speech discrimination scores for bisyllabic intelligibility:88.570%、85.714%、75.788%，speech discrimination scores for sentence intelligibility:89.785%、82.158%、88.600%.c.①Applying SAMs and MSPTMto CI subjects respectively have great impact on their speech recognition performance,and there is significantdifference in speech recognition performance (F=25.328，P=0.000，P﹤0.05).②When subjects are tested at quiet state, at state with SNR+15dB,and at state with SNR-5dB state respectively, there is significant differencein speech recognition performance (F=36.685，P=0.000，P﹤0.05).③Adopting SAMs and MSPTMrespectively and using differentspeech audiometry word lists have great impact on speech recognitionperformance and show significant difference (F=12.592，P=0.000, P﹤0.05).④Adopting SAMs and MSPTMrespectively and when subjects aretested at quiet state, at state with SNR+15dB, and at state with SNR-5dBstate respectively, there is significant difference in speech recognitionperformance (F=16.114，P=0.000，P﹤0.05).Conclusion: A Chinese speech-signal test platform with man-machineinterface in complex listening environments was designed and set up, basedon which normal hearing in four-channel cochlear implant of simulationprocessing by Sichuanhua speech materials (SAMs), which are suitable forSichuanhua-speaking impaired hearing (HI) patients (including hearingaids (HAs) and cochlear implant (CI) patients) will be developed. Fromthree aspects are consistent and stable, which indicating that SAMs as a clinical dialect materials, available in sichuanhua speakers to the speechperception evaluation and training of rehabilitation; And the speech-signaltest platform with man-machine interface is reliability and validity offeasible and high sensitivity. And comparing those speech discriminationscores not only have offered reliable data for developing speech processorfor Chinese-speaking and Chinese-dialect speaking CI users and forimproving the performance of noise-reducing software a new researchdirection for developing other Chinese-dialect acoustic speech materials. |
PDF Full Text Request