• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2007년도 추계학술대회논문집
• /
• pp.1009-1013
• /
• 2007

To do a HRTF customization, researchers used a spherical head model for modeling the head block of structural modeling of HRTF, which is the one of the technique for HRTF customization, because of its simplicity. In this paper, an analytic spheroidal HRTF caused by an incident point source will be introduced. Using proposed spheroidal HRTF, near-field HRTF customization can be applicable through a structural modeling of HRTF. To see the necessity of sheroidal head model, comparison of two analytic solutions, which are classical spherical HRTF and proposed spheroidal HRTF, will be shown. On the view point of ITD, optimal head model which matches with the measured ITD of KEMAR HRTF can be obtained. ITD results show that there are only slight differences between spherical and spheroidal head model. Magnitude comparison is made by constructing head model using measured head size. Although magnitude comparison is not studied between optimal models, the results of 24 of 36 subjects are shown that spheroidal head model matches notch frequency pattern of measured HRTF better than those of spherical one, where the sound source is at contralateral position.

• 한국음향학회지
• /
• 제28권5호
• /
• pp.409-415
• /
• 2009

가상 음원의 공간화에 중요한 역할을 담당하는 머리전달 함수는 개인 별로 각기 다른 특성을 갖는다. 개인화된 머리 전달 함수를 얻기 위해서는 많은 시간과 특정한 장치를 필요로하며, 이에 따라 타인으로부터 얻어진 머리 전달 함수를 청취자의 특성에 맞도록 개인화 시키는 방법이 주로 사용된다. 본 논문에서는 머리 전달 함수의 개인화 방법으로 여러 개의 미리 측정된 머리 전달 함수 중에서 사용자에 가장 적합한 머리 전달 함수를 찾는 방법을 제안하였다. 제안된 방법은 청취자에게 적합한 머리 전달 함수를 빠르고 정확하게 찾기 위해 이진 탐색 트리를 이용한 분할 기법을 사용하였다. 분할 시 왜곡 척도는 사람이 귀에 방향감의 차이를 가장 잘 인지하는 왜곡 척도를 실험적으로 결정하였다. 캘리포니아 데이비스 대학에서 제공하는 CIPIC 머리 전달 함수 데이터 베이스를 이용하여 본 기법의 유용성을 평가하였다. 10명의 실험자가 참여한 청취 테스트에서 제안된 기법을 통해 탐색된 머리 전달 함수는 청취자 본래의 머리 전달 함수와 비교하여 가상 음원의 방향감에 있어서 매우 근접한 결과를 나타내었으며, 기존의 해부학적인 파라메터를 이용한 머리 전달 함수의 탐색 기법과 비교하여 우수한 성능을 나타내었다.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2009년도 춘계학술대회 논문집
• /
• pp.103-107
• /
• 2009

Because individual HRTFs (Head-Related Transfer Functions) vary from a person to a person, a HRTF database has been measured by researchers to investigate the inter-subject variation, and to generate high fidelity virtual sound image. Individual HRTFs not only vary between subjects but also vary due to wearing clothes and glasses in daily life. However, influence of different dressing condition on the measured HRTF was not sufficiently investigated. To quantify the effect of wearing clothes and glasses, dummy's HRTF is measured in an anechoic chamber with various dressing condition, and is evaluated in the sense of spectral distortion. HRTFs are measured both in the median plane and in the horizontal plane. In the median plane, under 6kHz, effect of different wearing clothes and glasses is negligible. Over 6kHz, however, effect of clothing distorts HRTF about 6dB in the sense of spectral distortion. Moreover, at high frequencies, effect of glasses is no longer negligible. In the horizontal plane, at some azimuths, even additional light cloth over the dummy can change the spectrum of HRTF (6dB spectral distortion) especially when sound source is at contralateral positions. Therefore, HRTF measurement with different wearing conditions can broaden the capability of HRTF customization whose technique utilizes a HRTF database.

• 한국소음진동공학회논문집
• /
• 제17권7호
• /
• pp.638-648
• /
• 2007

CIPIC HRTF database의 주성분 분석(PCA)을 통해 개인의 HRIR이 정규 직교화된 소수의 기저함수들의 선형 결합으로 잘 묘사됨을 알 수 있다. 이 기저함수들은 음원의 고도각, 청취자 마다 달라지는 HRIR의 변화를 표현할 수 있다. 선형결합에 사용되는 기저함수들의 가중치들은 음원의 고도각에 따라 특이한 경향을 지닌다. 또한, 각각의 음원 위치에서 가중치의 표준편차 크기순으로 기저함수의 중요도를 결정할 수 있다. 이 논문에서는 각 음원 위치마다 중요한 3개 기저함수의 가중치를 청취자가 직접 조절하게 함으로써 맞춤형 HRIR을 생성하는 방법을 제안한다. 주관평가 결과, 청취자의 음원 고도각 인지 성능과 음원 앞-뒤 구분 성능이 향상됨을 확인하였다.

• 한국음향학회지
• /
• 제30권4호
• /
• pp.167-180
• /
• 2011

Recently, 3-D audio technologies are commonly implemented through headphones. A major problem of the headphone-based 3-D audio is in-the-head localization, which occurs due to the inaccurate Head-Related Transfer Function (HRTF). Since the individual measurements of HRTFs are impractical, there have been several researches for HRTF customization. In this paper, an efficient method of customizing HRTFs for the sound externalization is proposed. Firstly, it is determined which part will be customized in HRTF through psychoacoustical experiments. Then, the method controlling spectral notches and envelopes to provide individual localization cues are described. Since the proposed method is based on a critical-band rate, the structure is much simpler than that of previous studies, but still effective. The performance was evaluated through a series of subjective tests, and the results confirmed that the customized HRTF using proposed method could replace the measured individual HRTF successfully.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2008년도 춘계학술대회논문집
• /
• pp.934-938
• /
• 2008

To customize individual characteristics of HRTF, a spherical model has been used for structural modeling technique. A pseudo-code of prolate spheroidal HRTF caused by incident acoustic point source is already developed, and it can be used a head shadow filter for structural modeling of HRTF. In this research, to see the necessity and efficiency of spheroidal head modeling, ITD optimization is performed on CIPIC HRTF database. From given cost function, ITD-optimized spheroidal head model, whose ITD information is the most matched version of measured ITD information, is found by varying head parameters subject by subject. By comparing results of ITD-optimized spheroids and ITD-optimized spheres, we concluded that a spherical head model is more efficient way of generating head shadow effect than a spheroidal head model does.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2005년도 추계학술대회논문집
• /
• pp.641-644
• /
• 2005

This study reveals some recent attempt in modeling empirically obtained B&K HATS (Head and Torso Simulator) HRTFs (Head Related Transfer Functions) to Isolate parameters that stimulate lateral and elevation perception. Localization using non-individual HRTFs often yields poor performance in synthesizing virtual sound sources when applied to a group of individuals due to differences in size and shape of head, pinnae, and torso. For realization of both effective and efficient virtual audio it is necessary to develop a method to tailor a given set of non-individual HRTFs to fit each listener without measuring his/her HRTF set. Pole-zero modeling is applied to fit HRIRs (Head Related Impulse Responses) and modeling criterions for determining suitable number of parameters are suggested for efficient modeling. Horizontal HRTFs are modeled as minimum-phase transfer functions with appropriate ITDs (Interaural Time Delay) obtained from RTF (Ray Tracing Formula) to better fit the size of listener's head for usage in simple virtualizer algorithms without complex regularization processes. Result of modeling HRTFs in the median plane is shown and parameters responsible for elevation perception are isolated which can be referred to in the future study of developing customizable HRTFs.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2007년도 춘계학술대회논문집
• /
• pp.120-126
• /
• 2007

A principal components analysis of the entire median HRIRs in the CIPIC HRTF database reveals that the individual HRIRs can be adequately reconstructed by a linear combination of several orthonormal basis functions. The basis functions cover the inter-individual and inter-elevation variations in median HRIRs. There are elevation-dependent tendencies in the weights of basis functions, and the basis functions can be ordered according to the magnitude of standard deviation of the weights at each elevation. We propose a HRIR customization method via tuning of the weights of 3 dominant basis functions corresponding to the 3 largest standard deviations at each elevation. Subjective listening test results show that both front-back reversal and vertical perception can be improved with the customized HRIRs.