Chan Jun Chun

Crosstalk Cancellation for Spatial Sound Reproduction in Portable Devices with Stereo Loudspeakers

To reproduce spatial sound through stereo loudspeakers in a portable device environment, it is important to properly design a crosstalk cancellation algorithm that cancels out acoustical crosstalk signals. In other words, the difference between the direct path of head-related transfer functions (HRTFs) and the crosstalk path of HRTFs is very small at certain frequencies, and this causes an excessive boost of frequencies when designing a crosstalk cancellation filter. To mitigate this problem, we propose a crosstalk cancellation filter design method that allows for the selective attenuation of unwanted peaks in the spectrum by constraining the magnitude of the difference between the direct and crosstalk path. The performance of the proposed method is evaluated by subjective source localization and objective tests. It is shown from the tests that the proposed method can provide improved spatial sound effects with very closely spaced stereo loudspeakers.

Sound source elevation using spectral notch filtering and directional band boosting in stereo loudspeaker reproduction

In this paper, a virtual elevation method of a sound source is proposed for a stereo loudspeaker reproduction. To achieve this goal, the proposed method first applies a head-related transfer function (HRTF) to the sound source to vertically localize it. Then, spectral notch filtering followed by directional band boosting is used for the further elevation of the sound source. In particular, a filter having three notches is designed by analyzing the spectral characteristics estimated from the measured HRTF database and by investigating the effect of spectral boosting on the perception of sound elevation. To evaluate the elevation performance of the proposed method, subjective listening tests are subsequently conducted using several sound sources, including male and female speech, as well as guitar, violin, and pop music. It is shown from the tests that the sound sources processed by the proposed method are most likely to be perceived as if they are played from stereo loudspeakers located at around 20°, even though the stereo loudspeakers are actually located on a horizontal plane, i.e.,0o.1.

Usability Evaluation of a User Interface for Mono-to-Stereo Audio Extension Applied to Portable Multimedia Devices

본 논문에서는 대부분의 휴대용 멀티미디어 기기들이 모노 마이크로폰을 탑재하거나, 스테레오 마 이크로폰을 인접 배치하여 거의 모노에 가까운 오디오 신호를 획득한다는 점을 감안하여 모노 오디 오를 스테레오 오디오로 확장하는 기법을 제안한다. 또한, 향후 사용자가 손쉽게 활용하고 조작 가능 하도록 사용자 인터페이스를 설계하고, 이에 대한 사용성을 평가한다. 구체적으로, 모노 오디오를 선 형 예측 분석을 통하여 오디오 특징 벡터를 추출하고, 이를 기존 스테레오 오디오 콘텐츠들을 통하여 기학습된 hidden Markov model을 활용하여 스테레오 오디오로 확장한다. 또한, 제안된 기법이 휴대 용 멀티미디어 기기에서 손쉽게 활용 가능하도록 어플케이션 환경에서의 user interface (UI)를 설계 하고, 총 50명의 사용자에게 그 사용성을 평가하였다. 사용성 평가 결과, 기존 모노 오디오에 비하여 78%의 사용자가 제안된 기법을 통해 확장된 스테레오 오디오를 선호하였으며, 86%의 사용자가 제안 된 기법의 UI에 대하여 편리하다고 평가하였다.

An integrated approach of 3D sound rendering techniques for sound externalization

In this paper, a sound externalization method is proposed for out-of the head localization in headphone listening environments. Several externalization methods have been proposed that use either a head-related transfer function (HRTF) or early reflections. However, such conventional methods have drawbacks, e.g., timbre distortion due to the use of a measured HRTF or reverberation. On the other hand, the proposed externalization method integrates a model-based HRTF with reverberation. In addition, for improving frontal externalization performance, techniques such as decorrelation and spectral notch filtering are included. To evaluate the performance of the proposed externalization method, subjective listening tests are conducted by using different kinds of sound sources such as white noise, sound effects, speech, and music samples. It is shown from the test results that the proposed externalization method can localize sound sources farther away from out of the head than conventional methods.

A Real-Time Audio Upmixing Method from Stereo to 7.1-Channel Audio

In this paper, we propose a new method of upmixing stereo signals into 7.1-channel signals in order to provide more auditory realism. The proposed upmixing method employs an adaptive panning and a decorrelation technique for making more channels and reproducing natural reverberant surround sounds, respectively. The performance of the proposed upmixing method is evaluated using a MUSHRA test and compared with those of conventional upmixing methods. It is shown from the tests that 7.1-channel audio signals upmixed by the proposed method are preferred, compared to not only their original stereo audio signals but also 7.1-channel audio signals upmixed by conventional methods.

Upmixing Stereo Audio into 5.1 Channel Audio for Improving Audio Realism

In this paper, we address issues associated with upmixing stereo audio into 5.1 channel audio in order to improve audio realism. First, we review four different upmixing methods, including a passive surround decoding method, a least-mean-square based upmixing method, a principal component analysis based upmixing method, and an adaptive panning method. After that, we implement a simulator that includes the upmixing methods and audio controls to play both stereo and upmixed 5.1 channel audio signals. Finally, we carry out a MUSHRA test to compare the quality of the upmixed 5.1 channel audio signals to that of the original stereo audio signal. It is shown from the test that the upmixed 5.1 channel audio signals generated by the four different upmixing methods are preferred to the original stereo audio signals.

Application of low-frequency ultrasonic communication to audio marker for augmented reality

In augmented reality (AR), image markers are widely used for rendering virtual objects. However, the performance for image markers strongly depends on the lighting environment and the distance between the image marker and a camera. Therefore, we propose an audio marker-based AR application using low frequency ultrasound (LFU) communication. The proposed method consisted of forward error correction (FEC) coding and windowed differential phase shift keying (DPSK) modulation in order to make the proposed method robust over the noisy channel. The successful transmission rate (STR) performance of the proposed system was measured according to the various distances and azimuths under noisy conditions. As a result, the proposed audio marker transceiver could operate up to around 4 m without transmission error, even at ambient noise of 70 dB.

Conversion of nearly monaural audio to 5.1-channel audio for portable multimedia devices

In this paper, an upmixing method is proposed which converts nearly monaural audio signals into 5.1-channel ones for portable multimedia devices. Since spatial information is not clear due to close spacing between stereo microphones in portable devices, the proposed method tries to estimate the source direction depending on the frequency using azimuth-frequency analysis instead of directly estimating the inter-channel time difference or the inter-channel level difference between the two microphones. Then, a frequency-dependent amplitude panning technique is proposed to generate 5.1-channel audio signals. The performance of the proposed method is evaluated by examining the directivity patterns of upmixed 5.1-channel audio signals. Consequently, it is shown that the proposed method provides better directivity patterns than a conventional method based on passive surround decoding.

Real-Time Implementation of an Artificial Stereo Audio Extension Method

In this paper, we address the issue of the real-time implementation of an artificial stereo audio extension method that provides stereophonic audio from a mono audio source. First, the floating-point arithmetic programming of the method is performed. Then, the computationally complex parts of the floating-point operations are converted into fixed-point operations to allow them to operate in real time. The performance of a real-time implementation is evaluated by measuring the processing time and root mean square (RMS) value. Consequently, it is shown that partial fixed-point arithmetic programming reduces the processing time by 48.7%, with an RMS value of-90.50 dB, compared to the floating-point arithmetic programming.

Perceptual Enhancement of Sound Field Reproduction in a Nearly Monaural Sensing System

In this paper, we propose a method for enhancing the sound field in a nearly monaural sensing system. The proposed method controls the inter-channel coherence (ICC) of a stereo sound source by incorporating early reflections into the sound source. The degree of ICC is controlled using both a decorrelator and a principal component analysis (PCA) technique, while a cascaded all-pass filter is used to induce early reflections. To demonstrate the effectiveness of the proposed method on the perception of the sound field, a subjective test is carried out. It is shown from the test that the proposed method can reduce the ICC, thus it improves the perceived sound field.