臺灣大學: 工程科學及海洋工程學研究所柯文俊韓欣儒Han, Shin-RuShin-RuHan2013-03-272018-06-282013-03-272018-06-282010http://ntur.lib.ntu.edu.tw//handle/246246/252522許多中度與重度聽力損失之患者，聽力圖都具有之高頻衰減之特性。這些患者能夠正常分辨低頻訊號，但是對於高頻音訊之分辨力則較弱。移頻助聽器乃是將輸入語音之高頻成分轉移至低頻區域；對於低頻殘存之重度或完全損失感音神經性聽障患者，應有良好之幫助。早期對於移頻技術之研究之結果大多不理想；且其設備於臨床上從未獲得良好之迴響。然而隨著助聽器之技術日益增進，移頻助聽器可將成為高頻聽力損失患者之另一項選擇。慢放是最早實現移頻功能於助聽器之方法，然而慢放會使輸出語音時長變長，造成輸入與輸出不同步。本文結合音長修正法與慢速播放法模擬語音移頻後之結果，音長修正法能夠在不改變語音頻譜特性之前提下，修正語音之音長或音高，使得移頻後之語音時長與輸入相同。本文使用三種演算法:最小平方短時間修正傅立葉轉換估計法、同步疊加法、時域基頻同部疊加法，做為本文之音長修正法，在分析的過程中透過時間歷時圖、時頻圖與音高軌跡圖對三種方法之模擬結果做出結論。最小平方短時間修正傅立葉轉換估計法之結果穩定但運算過於耗時，同步疊加法速度較快但其結果會受到音框長度之影響。時域基頻同部疊加法則對於語音中擦音與塞擦音的部分會有較不正確的修正結果，然而整體而言，三種方法中是以時域基頻同部疊加法之結果最好。Many listeners with medium-severe hearing loss present audiograms with high frequency loss descending profiles. These patients can resolve spectral cues normally for lower frequency signals, but they often show less ability to use high frequency information. Frequency shifting hearing aids, which transpose the high frequency components of input sounds to the low frequency region, are known to be beneficial for sensorineural severe-to-profound hearing impaired listeners with residual hearing in the low frequency region. Early studies with frequency-lowering technology reported mostly unfavorable results, and the devices never gained clinical popularity. However, as hearing aid technology becomes ever more sophisticated, it is appropriate to reconsider whether frequency-lowering hearing aids should be an amplification choice in those with high-frequency hearing loss. Slow playback is the earliest method of implementing frequency-lowering in hearing aids, while resulting in longer output signals. This can cause the input and output signals to be desynchronized in time. In this research, time-scale modification and slow playback are combined to simulate frequency-shifted speech. Time-scale modification can change signal length without changing the spectrum characters and let output signal the same length with input. The time-scaling algorithms used in this paper are LSEE-MSTFTM, SOLA and TD-PSOLA. In the analysis process, the time frequency analysis diagram and the pitch contour diagram are utilized to compare simulation result. G&L method has more steady result, but it costs too much time. SOLA method is faster while the frame length will affect the output result. For the affricatives in speech ,TD-PSOLA method tend to has incorrect speech length, On the whole, TD-PSOLA method has the best result in three methods.9321168 bytesapplication/pdfen-US移頻助聽器音長修正法最小平方短時間修正傅立葉轉換估計法同步疊加法時域基頻同部疊加法Frequency shifting hearing aidsTime-scale modificationLSEE-MSTFTMSOLATD-PSOLA[SDGs]SDG3thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/252522/1/ntu-99-R97525031-1.pdf