以非線性訊號分析法探討人耳處理音樂訊號中包封的振幅變化之認知神經機制;A Nonlinear Signal Processing Approach to Investigate the Perceptual and Neural Correlates Underlying Processing the Temporal Envelope-Modulation Dimension of Sounds

NCUIR > college of Health Sciences and Technology > College of Science Institute of Cognitive Neuroscience > Research Project > Item 987654321/82364

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/82364

Title:	以非線性訊號分析法探討人耳處理音樂訊號中包封的振幅變化之認知神經機制;A Nonlinear Signal Processing Approach to Investigate the Perceptual and Neural Correlates Underlying Processing the Temporal Envelope-Modulation Dimension of Sounds
Authors:	謝宜蕙
Contributors:	國立中央大學認知與神經科學研究所
Keywords:	振幅調變;包封;非線性訊號演算法;時間週期序列;聽覺皮質區;amplitude modulation;envelope;nonlinear decomposition method;periodotopy;auditory cortex
Date:	2020-01-13
Issue Date:	2020-01-13 14:48:06 (UTC+8)
Publisher:	科技部
Abstract:	所有自然界的聲音皆是由一個隨著時間非線性變動的振幅調變(amplitude modulation; AM)包封(envelope)去調節一載波(carrier)而組成，不同的振幅調變速率(AM rate)提供了我們解析語言、音樂、和聲音場景分析的重要訊息。由於目前探討聲音振幅調變的研究皆以正弦波的包封為主，並且倚賴主流的線性分析方法如傅立葉轉換（Fourier Transform）。因考量自然界的聲音訊號多為非穩態或非線性，在假設線性的情況下，無法解析到瞬時的包封變化對人耳感知聲音的影響。另一方面，在大腦聽覺皮質區是否有處理振幅調變時間週期所相對應的腦區目前文獻上仍有爭議。本計畫目的為利用希爾伯特黃非線性訊號演算法(Hilbert-Huang Transform)探討人耳處理振幅調變包封的相關機制。實驗一分別用音調識別試驗和頻率閾限追蹤法探討訊號中包封調變速率、波型和載波頻率之關係對人耳感知音頻之影響，並將其結果與線性和非線性模型模擬出來的比較。實驗二則利用訊號處理方式創造一組消失的基頻音樂訊號(missing-fundamental harmonic complex)，並將其相同的包封移到高頻載波，藉以解離時頻域資訊對人耳分析振幅調變訊號的影響，此部分實驗有別於以往研究，利用絕對音感受試者可直接判定此訊號的音符名稱。實驗三以稀疏時間取樣(sparse-sampling)進行聽覺功能性磁振造影，操弄不同調變速率包封的載波和寬頻雜訊(AM broadband noise)來探討聽覺皮質區是否針對漸增的振幅調變速率呈現相對應的腦神經區域，預期會在聽覺皮質區呈現對應振幅調變時間週期序列(periodotopy)式的反應區域。 ;All natural sounds are composed of dynamic nonlinear temporal modulations of acoustic features (envelope) on carrier signals. The different time scales of the envelope modulation provides the most informative cues in music, speech perception, and auditory scene analysis. However, previous studies on temporal modulation have mostly used tones with simple sinusoidal modulations based on conventional spectral analyses grounded on data linearity, leaving the role of nonlinear envelope-modulation unexplored. At the cortical level, inconsistent findings exist regarding whether the rate of temporal modulation of sounds forms the fundamental organization principle at human auditory cortex. The proposed research aim to determine the perceptual response and neural correlates of human auditory cortex underlying processing of the temporal envelope dimension of amplitude-modulated (AM) signals by applying a nonlinear decomposition analysis method. Experiment 1 investigates pitch extraction from AM two-tone complexes with different combinations of carrier and envelope modulation patterns. Two pitch-matching and pitch- discrimination tasks will be used to test the ability of listeners to estimate pitch and discriminate small frequency differences produced by AM and two-tone signals. Pitch performance accuracy will be compared to model predictions based on conventional spectral analysis (e.g. Fourier transform) and nonlinear decomposition method (Ensemble Empirical Mode Decomposition with Hilbert-Huang transform). Experiment 2 examines whether pitch extraction from temporal modulation pattern is contingent on spectral region of the carrier signal. A novel transposed harmonic complex stimuli is developed to dissociate the tone’s spectral and temporal modulation information. Absolute-pitch musicians will be used to identify the musical note-name associated with the missing-fundamental frequency of a harmonic complex at low-frequency and the same modulation pattern transposed to high-frequency carriers. It is expected that pitch percepts corresponding to high-frequency transposed tones will significantly diminish if pitch extraction mechanism based on temporal modulation is insufficient for pitch coding. The last part of the proposed research aim to determine whether the rates of temporal modulation (periodotopy) drives the primary organization across human auditory cortex. A sparse-sampling paradigm in functional magnetic resonance imaging (fMRI) will be employed to characterize neural responses underlying envelope modulations of different rates and structures using AM sounds and broadband noises. It is hypothesized that auditory cortex would exhibit“periodotopic" organization dependent on the rate of temporal envelope modulations of sounds in primary auditory cortex and surrounding auditory subregions.
Relation:	財團法人國家實驗研究院科技政策研究與資訊中心
Appears in Collections:	[College of Science Institute of Cognitive Neuroscience] Research Project

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	253	View/Open

社群 sharing

Loading...