由於綠色能源議題在國際上的重要性越來越受重視，因此近年來台灣西部沿
海預計有大量的離岸風機基樁架設及運作，其對於海洋生態環境等危害更不可忽
視，持續性的低頻環境噪音對於海洋哺乳類等生物可能會造成嚴重的聽力受損甚
至危害其生命，鯨豚類的活動在風機打樁施工時的即時監測尤為重要。本研究擬
利用被動聲學監測搭配梅爾倒頻譜係數進行海洋鯨豚類哨叫聲聲紋辨識，其特徵
參數擷取法之頻帶分布相似於人耳聽力系統非線性特性，在低頻的解析度較高，
並且大量應用於語音聲紋辨識技術當中。
本研究之目的除辨識目標訊號以外，所接收之聲學資料皆包含訊號走時差，
也就是目標訊號經由不同路徑傳遞抵達後產生時間上的延遲差異，而在不同空間
下之水下音傳接收資料均涵蓋因多重路徑效應導致之相異延遲時間，透過掌握各
空間下之水中脈衝通道響應並建置資料庫後，即可利用聲紋辨識達到聲源定距之
目的。根據主動聲學實驗資料及結果分析發現，以三種不同聲學訊號拍發下，聲
紋辨識成功率達 97.86%，而於南海進行之聲源定距實驗，在聲源與接收器距離
兩公里以內之定距誤差在 0.5 公里以內，因此梅爾倒頻譜特徵參數擷取法可有效
運用於水中目標物之聲紋辨識，並且聲學資料中走時差特徵亦可作為聲源定位之
特徵，透過辨識與比對可有效定位聲源與接收器間距離。

With the development of renewable energy, increasingly importance has been
attached to offshore wind power. Accordingly, there are numerous foundation pile
erected and operated at the west shore of Taiwan. The harm to the marine ecological
environment should not be neglected. The constantly low-frequency background noise
could cause severe hearing impairment. So that the real-time monitoring to dolphin
activities when the construction of foundation pile plays a vital role. The purpose of
this study was to investigate the voiceprint recognition of the whistle of dolphin using
Mel-frequency Cepstrum Coefficient. The frequency band distribution of feature
extraction is similar to human auditory system. The hearing resolution of low-frequency
is higher than high-frequency. Nowadays, it is widely applied to speech recognition.
Another aim was to use the delay time of the received acoustic data to ranging the
source from the receiver. Because of the propagation medium, the signals transmission
is limited to the sea surface and seabed. The different path reflection by boundary forms
the unique characteristic of delay time to different spatial distribution of the source. By
the database creation of different spatial acoustic data with impulse response, we can
therefore use the voiceprint recognition to range the distance between source and
receiver. It was found that from the received data by the three different type of the
source signal transmitted in water, the accuracy of voiceprint recognition was about
97.86%. And the source ranging experiment implemented at the South China Sea, the
ranging error distance was less than 0.5 km when the source and receiver near than 2
km. The results revealed that the proposed method is very promising for classification
of underwater transient signals. And the characteristic of delay time to the received data
can be the feature of source ranging as well.

論文審定書............................................................................................................. i
謝誌........................................................................................................................ ii
摘要....................................................................................................................... iii
Abstract ................................................................................................................. iv
目錄....................................................................................................................... vi
圖次..................................................................................................................... viii
表次...................................................................................................................... xii
第一章 緒論.................................................................................................... 1
1.1 前言............................................................................................. 1
1.2 研究動機及目的......................................................................... 2
1.3 文獻回顧..................................................................................... 4
1.4 論文架構................................................................................... 10
第二章 訊號處理與數值方法...................................................................... 11
2.1 帶通濾波（Band-pass Filter） ................................................ 11
2.2 調變及重新採樣....................................................................... 12
2.3 端點偵測 (End-Point Detection) ............................................. 14
2.4 特徵參數擷取........................................................................... 22
2.5 聲紋比對與辨識....................................................................... 26
2.6 接收聲學資料能量分析........................................................... 27
第三章 高雄港外水下目標物聲紋辨識實驗.............................................. 28
3.1 實驗使用儀器介紹................................................................... 29
3.2 實驗使用之聲學訊號............................................................... 36
3.3 實驗資料分析結果................................................................... 38
3.4 小結........................................................................................... 43
第四章 南海海域水下目標物偵測與聲源定位實驗.................................. 44
4.1 實驗使用儀器介紹................................................................... 46
4.2 實驗使用之聲學訊號............................................................... 51
4.3 聲紋資料庫建置....................................................................... 53
4.4 南海實驗資料分析結果........................................................... 62
第五章 結論.................................................................................................. 76
參考文獻............................................................................................................. 78

[1] 經濟部能源局風力發電單一服務窗口(2019)，
https://www.twtpo.org.tw/gas.aspx?id=3268。
[2] 王亭勻(2009)。中華白海豚之保育策略。成功大學海洋科技與事務研究所
學位論文，1-115。
[3] 蔡秉及，連光山，林茂，林玉輝(1994)。廈門港及鄰近海域浮游動物的生
態研究，海洋學報，4。
[4] Yang, C. K. (2017)。船舶噪音對台灣西海岸中華白海豚之潛在影響。臺灣
大學生態學與演化生物學研究所學位論文，1-81。
[5] Thomsen, F., Lüdemann, K., Kafemann, R., & Piper, W. (2006). Effects of
offshore wind farm noise on marine mammals and fish. Biola, Hamburg,
Germany on behalf of COWRIE Ltd, 62.
[6] Caltrans. (2001). Pile Installation Demonstration Project, Fisheries Impact
Assessment. PIDP EA 012081, Caltrans Contract 04A0148.
[7] 周蓮香及李政諦(2010)。中華白海豚棲地熱點評估及整體保育方案規劃，
行政院農業委員會林務局。
[8] Nachtigall, P. E., Pawloski, J. L., & Au, W. W. (2003). Temporary threshold
shifts and recovery following noise exposure in the Atlantic bottlenosed
dolphin (Tursiops truncatus). The Journal of the Acoustical Society of
America, 113(6), 3425-3429.
[9] Nister, D., & Stewenius, H. (2006, June). Scalable recognition with a
vocabulary tree. In 2006 IEEE Computer Society Conference on Computer
Vision and Pattern Recognition (CVPR'06) (Vol. 2, pp. 2161-2168). Ieee.
[10] Tzanetakis, G., & Cook, P. (2002). Musical genre classification of audio
signals. IEEE Transactions on speech and audio processing, 10(5), 293-302.
[11] Huang, C. J., Yang, Y. J., Yang, D. X., & Chen, Y. J. (2009). Frog classification
using machine learning techniques. Expert Systems with Applications, 36(2),
3737-3743.
[12] Fagerlund, S. (2007). Bird species recognition using support vector
machines. EURASIP Journal on Advances in Signal Processing, 2007(1),
038637.
[13] Boashash, B., & O'shea, P. (1990). A methodology for detection and
classification of some underwater acoustic signals using time-frequency
analysis techniques. IEEE Transactions on Acoustics, Speech, and Signal
Processing, 38(11), 1829-1841.
[14] Tucker, S., & Brown, G. J. (2005). Classification of transient sonar sounds
using perceptually motivated features. IEEE Journal of Oceanic
Engineering, 30(3), 588-600.
[15] Malfante, M., Mars, J. I., Dalla Mura, M., & Gervaise, C. (2018). Automatic
fish sounds classification. The Journal of the Acoustical Society of
America, 143(5), 2834-2846.
[16] Atal, B., & Rabiner, L. (1976). A pattern recognition approach to voiced
unvoiced-silence classification with applications to speech recognition. IEEE
Transactions on Acoustics, Speech, and Signal Processing, 24(3), 201-212.
[17] Qiao, G., Ma, T., Liu, S., Zheng, N., Babar, Z., & Yin, Y. (2019, June). Spectral
Entropy Based Dolphin Whistle Detection Algorithm and Its Possible
Application for Biologically Inspired Communication. In OCEANS 2019
Marseille (pp. 1-6). IEEE.
[18] Erbe, C., & King, A. R. (2008). Automatic detection of marine mammals using
information entropy. The Journal of the Acoustical Society of America, 124(5),
2833-2840.
[19] Shen, J. L., Hung, J. W., & Lee, L. S. (1998). Robust entropy-based endpoint
detection for speech recognition in noisy environments. In Fifth international
conference on spoken language processing.
[20] Hsieh, S. C., Chen, W. P., Lin, W. C., Chou, F. S., & Lai, J. R. (2012). A Study
of the Application of an Average Energy Entropy Method for the Endpoint
Extraction of Frog Croak Syllables. 臺灣林業科學, 27(2), 177-189.
[21] Boashash, B., & Black, P. (1987). An efficient real-time implementation of the
Wigner-Ville distribution. IEEE transactions on acoustics, speech, and signal
processing, 35(11), 1611-1618.
[22] Atal, B. S. (1974). Effectiveness of linear prediction characteristics of the
speech wave for automatic speaker identification and verification. the Journal
of the Acoustical Society of America, 55(6), 1304-1312.
[23] Davis, S., & Mermelstein, P. (1980). Comparison of parametric representations
for monosyllabic word recognition in continuously spoken sentences. IEEE
transactions on acoustics, speech, and signal processing, 28(4), 357-366.
[24] Wang, W., Li, S., Yang, J., Liu, Z., & Zhou, W. (2016, January). Feature
extraction of underwater target in auditory sensation area based on MFCC.
In 2016 IEEE/OES China Ocean Acoustics (COA) (pp. 1-6). IEEE.
[25] Taegyun Lim，Keunsung Bae，Chansik Hwang，Hyeonguk Lee，
CLASSIFICATION OF UNDERWATER TRANSIENT SIGNALS USING
MFCC FEATURE VECTOR，IEEE，2007
[26] Friedlander, B. (1988). Accuracy of source localization using multipath
delays. IEEE Transactions on Aerospace and Electronic Systems, 24(4), 346
359.
[27] Sakoe, H., & Chiba, S. (1978). Dynamic programming algorithm optimization
for spoken word recognition. IEEE transactions on acoustics, speech, and
signal processing, 26(1), 43-49.
[28] Myers, C., Rabiner, L., & Rosenberg, A. (1980). Performance tradeoffs in
dynamic time warping algorithms for isolated word recognition. IEEE
Transactions on Acoustics, Speech, and Signal Processing, 28(6), 623-635.