Search

In this paper, we propose an online adaptation method for Fourier series-based acoustic transfer function (FS-ATF) models for robot audition systems using microphone array signal processing. The ATF represents the characteristics of signal propagation from a sound source to a microphone, which is essential for sound source localization (SSL) and sound source separation (SSS). For practical applications in dynamically changing real-world environments, ATF models must satisfy two criteria: (1) they must be adaptable to changes in the acoustic environment; and (2) they must be lightweight to be suitable for resource-constrained systems, such as robots with limited memory and computational capacity. The proposed method addresses these challenges using Fourier series expansions for interpolation, which reduces the memory footprint of the ATF model and facilitates online adaptation to acoustic environmental changes. The experimental results demonstrate that the proposed online adaptation method both improves the SSL and SSS performance while reducing the size of the ATF model, which represents a significant improvement over existing online ATF adaptation methods. [ABSTRACT FROM AUTHOR]

This paper proposes an environmental sound segmentation method using Mask U-Net. In recent years, human–robot interactions, especially speech dialogue, have been assessed by auditory scene analysis. Methods, such as noise reduction, section detection, and sound source separation have been proposed for robot audition, acoustic signal processing, and machine learning. However, such conventional approaches have three drawbacks: (1) Many studies have analyzed individual functions, which are regarded as being a cascade. Cascade systems can, however, result in the accumulation of errors generated at each functional block. (2) Unlike conventional cascade systems, deep learning-based methods that simultaneously detect sections, separate sound sources, and identify classes have also been proposed for speech separation. These techniques can be extended for multiple classes of environmental sounds, but their performance becomes degraded with large variations in sound event lengths among classes. (3) In addition, these methods have recurrent neural network layers making it difficult to process calculations in parallel. This paper proposes an environmental sound segmentation method called Mask U-Net, which robustly differentiates sound event lengths among classes. Simulation experiments using a developed 75-class environmental sound data set showed that the proposed method was faster than conventional methods and showed high segmentation performance. [ABSTRACT FROM AUTHOR]

Drone audition, drone's auditory capabilities for a multi-rotor helicopter (hereinafter, drone), has been developed to improve real-world tasks, e.g. search-and-rescue tasks, by compensating for the weakness of visual capabilities due to darkness and occlusion. Most of current implementations of robot audition focus on a single sound source. This paper focuses on the estimation of multiple sound source positions from acoustic signals captured by a drone equipped with a microphone array. Due to ego-noise such as rotor and airflow noise around the drone, the estimation of the sound source position is obscured and prone to error. In particular, in case of multiple sound sources, data association between localization information and sound sources is critical to the performance of such estimation. To cope with uncertainty in data association, we extend Global Nearest Neighbor (GNN) to exploit sound source features (GNN-c) because drone audition needs real-time or least latency. The resulting system demonstrates that it can estimate multiple sound source positions with an accuracy of about 3 m. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of 2D sound source localization using multiple microphone arrays in an outdoor environment. Two main issues exist in such localization. Since the localization performance depends on a variety of parameters, the lack of knowledge about how to design the system is one of those issues. A thorough analysis in respect to the accuracy of the localization results with different simulation conditions has been performed. Obtained characteristics lead to a discussion on limitations and applicability of the system. The distinction between multiple simultaneous sound sources is another problem. This is directly related to the appearance of outliers in the localization process. To solve this issue, an outlier removal method is proposed, which takes the properties of the observed sounds into consideration. In this paper a VR-based visualization method of the obtained results is also introduced. As the application scenario, we selected bird song analysis, which provides a challenging environment in terms of constantly changing signal-to-noise ratio and relative sensor-to-target position. A prototype system has been established using the proposed method. Several simulation results have been presented followed by a discussion on the issues. This leads to establishing system design guidelines that ensure a predictable performance. [ABSTRACT FROM AUTHOR]

This paper presents an interactive quizmaster robot that can manage a multiparty speech-based quiz game. The basic flow of the quiz game is that (1) the robot reads a question, (2) one or more players answer it, and (3) the robot judges the correctness of the answers. We categorize such speech-based quiz games into school-type interaction and auction-type interaction. The former asks players to say ‘Yes’ to get the right to answer before answering a question and the latter allows players to directly answer a question without any advance notice. To realize such interaction, the robot needs the capability of recognizing utterances from multiple people using its own microphones (i.e. ears), even if those utterances are made simultaneously. To cope with such situations, the robot estimates which player made the fastest utterance and recognizes it by localizing and separating a mixture of audio signals. Experiments were conducted to evaluate the success rates of the fastest player identification and speech recognition. The results showed that the robot could identify the fastest speakers with a success rate of 90.0% more accurately than humans when only one speaker slightly preceded the other speakers. We found that although the success rate of speech recognition for the fastest speakers did not reach that of humans, the robot attains amusing quiz game interaction. [ABSTRACT FROM AUTHOR]

Hose-shaped rescue robots have been developed for searching narrow spaces such as under collapsed buildings. The posture estimation independent of the past history is critical, because conventional inertial-sensor-based posture estimation has two main problems; a cumulative error problem peculiar to inertial sensors, and a sudden posture change problem caused by external forces. For coping with the two problems, we developed a novel posture estimation method by putting an active microphone array, a set of microphones and loudspeakers, on the robot. The method calculates the time difference of arrival (TDOA) of the reference signal emitted from one loudspeaker, and estimates the posture from the distance obtained by TDOA. This concise method still has three problems: (1) external noise, (2) reverberation and reflection, and (3) obstacles. These problems are tackled by (1) TSP signal, (2) GCC-PHAT and a threshold-based onset detection, and (3) rejecting incorrect onset times, respectively. Experiments with simulated sounds and actual recordings demonstrate that the method attains the performance of estimation comparable to that of conventional methods, that is, less than 20 cm of the tip position error. Even without historical data, the method attains the similar performance while conventional methods fail. [ABSTRACT FROM AUTHOR]

This study addresses a framework for a robot audition system, including sound source localization (SSL) and sound source separation (SSS), that can robustly recognize simultaneous speeches in a real environment. Because SSL estimates not only the location of speakers but also the number of speakers, such a robust framework is essential for simultaneous speech recognition. Moreover, improvement in the performance of SSS is crucial for simultaneous speech recognition because the robot has to recognize the individual source of speeches. For simultaneous speech recognition, current robot audition systems mainly require noise-robustness, high resolution, and real-time implementation. Multiple signal classification (MUSIC) based on standard Eigenvalue decomposition (SEVD) and Geometric-constrained high-order decorrelation-based source separation (GHDSS) are techniques utilizing microphone array processing, which are used for SSL and SSS, respectively. To enhance SSL robustness against noise while detecting simultaneous speeches, we improved SEVD-MUSIC by incorporating generalized Eigenvalue decomposition (GEVD). However, GEVD-based MUSIC (GEVD-MUSIC) and GHDSS mainly have two issues: (1) the resolution of pre-measured transfer functions (TFs) determines the resolution of SSL and SSS and (2) their computational cost is expensive for real-time processing. For the first issue, we propose a TF-interpolation method integrating time-domain-based and frequency-domain-based interpolation. The interpolation achieves super-resolution robot audition, which has a higher resolution than that of the pre-measured TFs. For the second issue, we propose two methods for SSL: MUSIC based on generalized singular value decomposition (GSVD-MUSIC) and hierarchical SSL (H-SSL). GSVD-MUSIC drastically reduces the computational cost while maintaining noise-robustness for localization. In addition, H-SSL reduces the computational cost by introducing a hierarchical search algorithm instead of using a greedy search for localization. These techniques are integrated into a robot audition system using a robot-embedded microphone array. The preliminary experiments for each technique showed the following: (1) The proposed interpolation achieved approximately 1-degree resolution although the TFs are only at 30-degree intervals in both SSL and SSS; (2) GSVD-MUSIC attained 46.4 and 40.6% of the computational cost compared to that of SEVD-MUSIC and GEVD-MUSIC, respectively; (3) H-SSL reduced 71.7% of the computational cost to localize a single speaker. Finally, the robot audition system, including super-resolution SSL and SSS, is applied to robustly recognize four sources of speech occurring simultaneously in a real environment. The proposed system showed considerable performance improvements of up to 7% for the average word correct rate during simultaneous speech recognition, especially when the TFs were at more than 30-degree intervals. [ABSTRACT FROM AUTHOR]

This paper addresses online calibration of an asynchronous microphone array. Although microphone array techniques are effective for sound localization and separation, these techniques have two issues; geometry information on a microphone array or time-consuming measurements of transfer functions between a microphone array and a sound source is necessary, and a fully synchronous multichannel analog-to-digital converter should be used. To solve these issues, we proposed an online framework for microphone array calibration by combining simultaneous localization and mapping (SLAM), and beamforming. SLAM simultaneously calibrates locations of microphones and a sound source, and clocks differences between microphones every time a microphone array observes a sound event. Beamforming works as a cost function to decide the convergence of calibration by localizing the sound using the transfer functions calculated from the estimated microphone locations and clock differences. We implemented a prototype system based on the proposed framework using extended Kalman filter-based SLAM and delay-and-sum beamforming. The experimental results showed that the proposed framework successfully calibrated an eight-channel asynchronous microphone array both in a simulated and a real environment even when system parameters such as variances are set to be 10 times larger than the optimal values. Furthermore, the error of sound localization with the calibrated microphone array was as small as the desired one, that is, the grid size for beamforming. [ABSTRACT FROM PUBLISHER]

This paper describes improvements in Audio-Visual Voice Activity Detection (AV-VAD) using a state transition model. Audio-Visual integration is a promising approach to improve the noise-robustness of VAD. Our proposed AV-VAD is based on a state transition model with four utterance states: speech, nonspeech, and two asynchronous states, a beginning motion state and an ending motion state, corresponding to lip activity before and after voice activity, respectively. We implanted a prototype system into an upper torso humanoid SIG and evaluated the ability of the proposed method by using auditory- and/or visually-contaminated data. Experimental results showed that the robustness of the VAD improved even when the resolution of images was low. On average, our proposed method resulted in a 7.8 point improvement in word detection rates compared with existing AV-VAD methods. [ABSTRACT FROM PUBLISHER]

The motors of a robot produce ego-motion noise that degrades the quality of recorded sounds. This paper describes an architecture that enhances the capability of a robot to perform automatic speech recognition (ASR) even as the entire body of the robot moves. The architecture consists of three blocks: (i) a multichannel noise reduction block, consisting of microphone-array-based sound localization, geometric source separation and post-filtering, (ii) a single-channel template subtraction block and (iii) an ASR block. As the first step of our analysis strategy, we divided the whole-body motion noise problem into three subdomains of arm, leg and head motion noise, according to their intensity levels and spatial location. Subsequently, by following a synthesis-by-analysis approach, we determined the best method for suppressing each type of ego-motion noise. Finally, we proposed to utilize a control module in our ASR framework; this module was designed to make decisions based on instantaneously detected motions, allowing it to switch to the most appropriate method for the current type of noise. This proposed system resulted in improvements of up to 50 points in word correct rates compared with results obtained by single microphone recognition of arm, leg and head motions. [ABSTRACT FROM AUTHOR]

This paper presents the design and implementation of the HARK robot audition software system consisting of sound source localization modules, sound source separation modules and automatic speech recognition modules of separated speech signals that works on any robot with any microphone configuration. Since a robot with ears may be deployed to various auditory environments, the robot audition system should provide an easy way to adapt to them. HARK provides a set of modules to cope with various auditory environments by using an open-sourced middleware, FlowDesigner, and reduces the overheads of data transfer between modules. HARK has been open-sourced since April 2008. The resulting implementation of HARK with MUSIC-based sound source localization, GSS-based sound source separation and Missing Feature Theory-based automatic speech recognition on Honda ASIMO, SIG2 and Robovie R2 attains recognizing three simultaneous utterances with the delay of 1.9 s at the word correct rate of 80–90% for three speakers. [ABSTRACT FROM AUTHOR]

People usually talk face to face when they communicate with their partner. Therefore, in robot audition, the recognition of the front talker is critical for smooth interactions. This paper presents an enhanced speech detection method for a humanoid robot that can separate and recognize speech signals originating from the front even in noisy home environments. The robot audition system consists of a new type of voice activity detection (VAD) based on the complex spectrum circle centroid (CSCC) method and a maximum signal-to-noise ratio (SNR) beamformer. This VAD based on CSCC can classify speech signals that are retrieved at the frontal region of two microphones embedded on the robot. The system works in real-time without needing training filter coefficients given in advance even in a noisy environment (SNR > 0 dB). It can cope with speech noise generated from televisions and audio devices that does not originate from the center. Experiments using a humanoid robot, SIG2, with two microphones showed that our system enhanced extracted target speech signals more than 12 dB (SNR) and the success rate of automatic speech recognition for Japanese words was increased by about 17 points. [ABSTRACT FROM AUTHOR]