Gaussian Process Regression for Single-Channel Sound Source Localization System Based on Homomorphic Deconvolution.

To extract the phase information from multiple receivers, the conventional sound source localization system involves substantial complexity in software and hardware. Along with the algorithm complexity, the dedicated communication channel and individual analog-to-digital conversions prevent an increase in the system's capability due to feasibility. The previous study suggested and verified the single-channel sound source localization system, which aggregates the receivers on the single analog network for the single digital converter. This paper proposes the improved algorithm for the single-channel sound source localization system based on the Gaussian process regression with the novel feature extraction method. The proposed system consists of three computational stages: homomorphic deconvolution, feature extraction, and Gaussian process regression in cascade. The individual stages represent time delay extraction, data arrangement, and machine prediction, respectively. The optimal receiver configuration for the three-receiver structure is derived from the novel similarity matrix analysis based on the time delay pattern diversity. The simulations and experiments present precise predictions with proper model order and ensemble average length. The nonparametric method, with the rational quadratic kernel, shows consistent performance on trained angles. The Steiglitz-McBride model with the exponential kernel delivers the best predictions for trained and untrained angles with low bias and low variance in statistics.

Single-Channel Multiple-Receiver Sound Source Localization System with Homomorphic Deconvolution and Linear Regression.

The conventional sound source localization systems require the significant complexity because of multiple synchronized analog-to-digital conversion channels as well as the scalable algorithms. This paper proposes a single-channel sound localization system for transport with multiple receivers. The individual receivers are connected by the single analog microphone network which provides the superimposed signal over simple connectivity based on asynchronized analog circuit. The proposed system consists of two computational stages as homomorphic deconvolution and machine learning stage. A previous study has verified the performance of time-of-flight estimation by utilizing the non-parametric and parametric homomorphic deconvolution algorithms. This paper employs the linear regression with supervised learning for angle-of-arrival prediction. Among the circular configurations of receiver positions, the optimal location is selected for three-receiver structure based on the extensive simulations. The non-parametric method presents the consistent performance and Yule-Walker parametric algorithm indicates the least accuracy. The Steiglitz-McBride parametric algorithm delivers the best predictions with reduced model order as well as other parameter values. The experiments in the anechoic chamber demonstrate the accurate predictions in proper ensemble length and model order.

Parametric Estimations Based on Homomorphic Deconvolution for Time of Flight in Sound Source Localization System.

Vehicle-mounted sound source localization systems provide comprehensive information to improve driving conditions by monitoring the surroundings. The three-dimensional structure of vehicles hinders the omnidirectional sound localization system because of the long and uneven propagation. In the received signal, the flight times between microphones delivers the essential information to locate the sound source. This paper proposes a novel method to design a sound localization system based on the single analog microphone network. This article involves the flight time estimation for two microphones with non-parametric homomorphic deconvolution. The parametric methods are also suggested with Yule-walker, Prony, and Steiglitz-McBride algorithm to derive the coefficient values of the propagation model for flight time estimation. The non-parametric and Steiglitz-McBride method demonstrated significantly low bias and variance for 20 or higher ensemble average length. The Yule-walker and Prony algorithms showed gradually improved statistical performance for increased ensemble average length. Hence, the non-parametric and parametric homomorphic deconvolution well represent the flight time information. The derived non-parametric and parametric output with distinct length will serve as the featured information for a complete localization system based on machine learning or deep learning in future works.

Monaural Sound Localization Based on Reflective Structure and Homomorphic Deconvolution.

The asymmetric structure around the receiver provides a particular time delay for the specific incoming propagation. This paper designs a monaural sound localization system based on the reflective structure around the microphone. The reflective plates are placed to present the direction-wise time delay, which is naturally processed by convolutional operation with a sound source. The received signal is separated for estimating the dominant time delay by using homomorphic deconvolution, which utilizes the real cepstrum and inverse cepstrum sequentially to derive the propagation response's autocorrelation. Once the localization system accurately estimates the information, the time delay model computes the corresponding reflection for localization. Because of the structure limitation, two stages of the localization process perform the estimation procedure as range and angle. The software toolchain from propagation physics and algorithm simulation realizes the optimal 3D-printed structure. The acoustic experiments in the anechoic chamber denote that 79.0% of the study range data from the isotropic signal is properly detected by the response value, and 87.5% of the specific direction data from the study range signal is properly estimated by the response time. The product of both rates shows the overall hit rate to be 69.1%.

Near-Field Sound Localization Based on the Small Profile Monaural Structure.

The acoustic wave around a sound source in the near-field area presents unconventional properties in the temporal, spectral, and spatial domains due to the propagation mechanism. This paper investigates a near-field sound localizer in a small profile structure with a single microphone. The asymmetric structure around the microphone provides a distinctive spectral variation that can be recognized by the dedicated algorithm for directional localization. The physical structure consists of ten pipes of different lengths in a vertical fashion and rectangular wings positioned between the pipes in radial directions. The sound from an individual direction travels through the nearest open pipe, which generates the particular fundamental frequency according to the acoustic resonance. The Cepstral parameter is modified to evaluate the fundamental frequency. Once the system estimates the fundamental frequency of the received signal, the length of arrival and angle of arrival (AoA) are derived by the designed model. From an azimuthal distance of 3-15 cm from the outer body of the pipes, the extensive acoustic experiments with a 3D-printed structure show that the direct and side directions deliver average hit rates of 89% and 73%, respectively. The closer positions to the system demonstrate higher accuracy, and the overall hit rate performance is 78% up to 15 cm away from the structure body.

Monaural sound localization based on structure-induced acoustic resonance.

A physical structure such as a cylindrical pipe controls the propagated sound spectrum in a predictable way that can be used to localize the sound source. This paper designs a monaural sound localization system based on multiple pyramidal horns around a single microphone. The acoustic resonance within the horn provides a periodicity in the spectral domain known as the fundamental frequency which is inversely proportional to the radial horn length. Once the system accurately estimates the fundamental frequency, the horn length and corresponding angle can be derived by the relationship. The modified Cepstrum algorithm is employed to evaluate the fundamental frequency. In an anechoic chamber, localization experiments over azimuthal configuration show that up to 61% of the proper signal is recognized correctly with 30% misfire. With a speculated detection threshold, the system estimates direction 52% in positive-to-positive and 34% in negative-to-positive decision rate, on average.

Lightweight filter architecture for energy efficient mobile vehicle localization based on a distributed acoustic sensor network.

The generic properties of an acoustic signal provide numerous benefits for localization by applying energy-based methods over a deployed wireless sensor network (WSN). However, the signal generated by a stationary target utilizes a significant amount of bandwidth and power in the system without providing further position information. For vehicle localization, this paper proposes a novel proximity velocity vector estimator (PVVE) node architecture in order to capture the energy from a moving vehicle and reject the signal from motionless automobiles around the WSN node. A cascade structure between analog envelope detector and digital exponential smoothing filter presents the velocity vector-sensitive output with low analog circuit and digital computation complexity. The optimal parameters in the exponential smoothing filter are obtained by analytical and mathematical methods for maximum variation over the vehicle speed. For stationary targets, the derived simulation based on the acoustic field parameters demonstrates that the system significantly reduces the communication requirements with low complexity and can be expected to extend the operation time considerably.

Binaural sound localizer for azimuthal movement detection based on diffraction.

Sound localization can be realized by utilizing the physics of acoustics in various methods. This paper investigates a novel detection architecture for the azimuthal movement of sound source based on the interaural level difference (ILD) between two receivers. One of the microphones in the system is surrounded by barriers of various heights in order to cast the direction dependent diffraction of the incoming signal. The gradient analysis of the ILD between the structured and unstructured microphone demonstrates the rotation directions as clockwise, counter clockwise, and no rotation of the sound source. Acoustic experiments with different types of sound source over a wide range of target movements show that the average true positive and false positive rates are 67% and 16%, respectively. Spectral analysis demonstrates that the low frequency delivers decreased true and false positive rates and the high frequency presents increases of both rates, overall.

Design and analysis of ultrasonic monaural audio guiding device for the visually impaired.

The novel Audio Guiding Device (AGD) based on the ultrasonic, which is named as SonicID, has been developed in order to localize point of interest for the visually impaired. The SonicID requires the infrastructure of the transmitters for broadcasting the location information over the ultrasonic carrier. The user with ultrasonic headset receives the information with variable amplitude upon the location and direction of the user due to the ultrasonic characteristic and modulation method. This paper proposes the monaural headset form factor of the SonicID which improves the daily life of the beneficiary compare to the previous version which uses the both ears. Experimental results from SonicID, Bluetooth, and audible sound show that the SonicID demonstrates comparable localization performance to the audible sound with silence to others.