個人化HRTF試聽調整機制之改善

Improvements to Listening-Adjusting Mechanism for Personalized HRTF

呂其衡(Chi-heng Lu)

指導教授 ： 張嘉銘

大同大學/電機資訊學院/資訊工程學系所/碩士(2011年)

摘要

近年來3D立體音效技術廣泛運用在各種應用，要產生逼真的3D立體音效，就必
須探討音源定位的問題。由於每個人的外觀多有差異，例如軀幹、肩膀、頭型、耳
廓大小的不同，會造就每個人獨特的HRTF。HRTF包含了聲波從音源到達耳膜間
所有的頻率響應，要進行音源定位，就必需有精確的個人化HRTF。
利用其他測試者已量測妥的HRTF資料為基礎，以合成個人化的HRTF，需先
分析並分類這些樣本資料，利用分類之特徵頻率，挑選較接近受測者之個人化
HRTF的樣本，來減少受測者試聽測試的次數。
合成試聽測試樣本的HRTF半環繞音效，讓測試者聆聽這些合成之半環繞音效，
並替換、調整其中感覺有聲音發散或是偏差的角度，最後就可得到受測者的個人化
HRTF。利用這樣的系統，可以不經由複雜、危險的測試便可得到每個人獨特的
HRTF資料。
本論文改善了先前研究於樣本選擇、替換的隨機性，以個人的特徵頻率為判斷
依據，與樣本資料進行比對，簡化測量時間並提昇效率。同時利用各角度間的關聯
性來調整角度替換後會產生的偏差，以降低個人化HRTF的誤差。

關鍵字

HRTF ； 3D聲音 ； 個人化

並列摘要

In recent years, the 3D audio effect is widely used in applications. In order to
produce 3D audio effect as real, the sound source localization must be discussed. Because
person's exterior is different to others, for example the torso, the shoulder, the head, and
auricle. These differences accomplish each person's unique HRTF. All the effects of
audio waves from sound source to eardrum are summarized and called as HRTF. A
precise personalized HRTF is required to produce a simulated sound at a specified
location.
To synthesize the personalized HRTF using the measured HRTF of others, these
measured data are analyzed and classified into several groups. The feature frequencies of
each group are used to decide which groups are more suitable for later test. Therefore, the
times of the listening-adjusting-cycles can be reduced.
Once the HRTF has been chosen, it is used to produce a simulated sound with
circular moving course. Then the listener listen to the simulated sound, and try to find the
location of sound source which is dispersed or error. After all the dispersed or error
locations are replaced and exchanged, the personalized HRTF is obtained without any
complex and risky measurement.
This paper improves the previous study which is based on the random sample selection.
The personal characteristics frequencies are used to determine which sample HRTF in
database is chosen. The time for measurement is saved, the efficiency for measurement is
improved. The relation between HRTF in each anger are used to adjust. Therefor, the
errors of HRTF is also reduced.

並列關鍵字

personalized ； 3D sound ； HRTF

參考文獻

of Head-Related Impulse Responses for Customized Spatial Audio,” WSEAS

[2] Alan Kraemer, “Two speakers are better than 5.1,” IEEE Spectrum, May, 2001.

[3] J. W. Strutt (Lord Rayleigh), “On our perception of sound direction,” Philosophical

[4] 張淑君,『空間音效模式化與數位實現之研究,』 國立成功大學，May 2002.

[5] Niklas Rober, Sven Andres and Maic Masuch, ”HRTF Simulations Through