TDOA在類神經網路與雙曲線之聲源定位比較分析

ASIA unversity > 資訊學院 > 資訊工程學系 > 博碩士論文 > Item 310904400/116963

Please use this identifier to cite or link to this item: http://asiair.asia.edu.tw/ir/handle/310904400/116963

Title:	TDOA在類神經網路與雙曲線之聲源定位比較分析
Other Titles:	Sound Source Localization Compare Analysis based on TDOA in Neural Networks and Hyperbola
Authors:	柯志欣 KE, ZHI-HSIN
Contributors:	施能義 SHIH, NENG-YIH 資訊工程學系
Keywords:	雙曲線;到達時間差;聲源定位;類神經網路 Hyperbolic;TDOA;Sound source localization;Neural network
Date:	2023
Issue Date:	2023-11-22 01:38:07 (UTC+0)
Abstract:	本研究旨在比較基於到達時間差（Time Difference of Arrival，TDOA）的聲源定位方法在類神經網絡和雙曲線定位算法上的性能特性。聲源定位是一項重要的議題，可以在聲源追蹤、噪音源判定、立體聲音處理和機器人導航等領域中發揮關鍵作用。本研究採用了類神經網絡和雙曲線兩種方法進行聲源定位，並在多個聲源下進行了對比分析。本研究架構設定距離1~10公尺和方向角15, 30, 45, 60, 75度的聲源數據集進行訓練和測試。通過比較類神經網絡和雙曲線方法在不同聲源位置變化下的定位精度，並評估了它們的應用差異。類神經網絡方法在某些情況下能夠實現較高的聲源定位精度，尤其在距離較遠的情況下表現出色。而雙曲線方法在距離較近的情況下表現良好，具有較好的實用性能。本研究提供了關於基於TDOA的聲源定位方法在類神經網絡和雙曲線定位算法上的比較分析。根據實際需求和應用場景，可以選擇合適的方法來實現準確且可靠的聲源定位。未來的研究可以進一步探索不同算法的優化和改進，並建立多模式混合系統，以提高聲源定位的性能和適用性。 The aim of this study is to compare the performance characteristics of sound source localization methods based on Time Difference of Arrival (TDOA) using two approaches: neural networks and hyperbolic curve-based algorithms. Sound source localization is an important topic that plays a crucial role in applications such as sound tracking, noise source identification, stereo sound processing, and robot navigation. This study adopts both neural network and hyperbolic curve methods for sound source localization and performs comparative analysis across multiple sound sources.The study framework includes training and testing datasets of sound sources at distances of 1-10 meters and direction angles of 15, 30, 45, 60 and 75 degrees. By comparing the localization accuracy of the neural network and hyperbolic curve methods under different variations of sound source positions, their differences in applicability are evaluated.The neural network method achieves higher sound source localization accuracy in certain scenarios, particularly at longer distances, exhibiting excellent performance. On the other hand, the hyperbolic curve method performs well in proximity situations and demonstrates better practical performance.This study provides a comparative analysis of sound source localization methods based on TDOA using neural networks and hyperbolic curve. Depending on the specific requirements and application scenarios, suitable method can be chosen to achieve accurate and reliable sound source localization. Future research can further explore optimization and improvements in different algorithms and establish multi-mode hybrid systems to enhance the performance and applicability of sound source localization.
Appears in Collections:	[資訊工程學系] 博碩士論文

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	13	View/Open

Loading...