Arthur P. Lobo

Voiced/unvoiced speech discrimination in noise using Gabor atomic decomposition

A new algorithm is developed for voiced-unvoiced speech discrimination in noise. Short segments of speech are modeled as a sum of basis functions from a Gabor dictionary. In each iteration, a Gabor atom is fitted (using the matching pursuit algorithm) to the residual obtained by subtracting the best-fit Gabor atom from the previous residual. Multiple discriminant analysis is used to reduce the dimensionality of the vector of Gabor coefficients to give a low-dimensional feature vector for classification. A radial basis function neural network is trained on the reduced feature vector set to discriminate between voiced and unvoiced speech/silence segments. On a database of 62 sentences in 5-dB SNR speech-shaped noise, 84% correct classification accuracy was obtained.

A PDA-based Research Platform for Cochlear Implants

Currently researchers interested in developing new signal processing algorithms for commercially available cochlear implants must rely on coding these algorithms in low-level assembly language. We propose a personal digital assistant (PDA) based research platform for developing and testing in real-time new signal processing strategies for cochlear implants. Software development can be done either in C or in LabVIEW. The C implementation can be further optimized using Intels primitive routines. In this paper, we report on the real-time implementation of a 16-channel noise-band vocoder algorithm, which is a similar algorithm used in commercially available implant processors. We further report on EEG recordings on the PDA acquired through a compact-flash data acquisition card.

A PDA platform for offline processing and streaming of stimuli for cochlear implant research

A PDA-based research platform has been developed for implementing novel speech processing strategies and conducting psychophysical experiments with cochlear implant (CI) research that do not necessarily require real-time processing. The developed interface streams stimuli pulses to a CI unit in an offline mode from a Personal Computer via PDA platform using Windows Sockets (WINSOCK). Front-end of the application is run in MATLAB where stimuli pulses are created. Winsock establishes a TCP/IP connection with the PDA and starts the transmission of stimuli data. Server application installed on the PDA reads the stimulation data and forwards it to the SDIO board in packets where it is forwarded to the cochlear implant unit and pulses are then played in realtime. Versatility and flexibility are the key characteristics of the platform for easy implementation and testing of a wide range of applications and experiments without advanced programming skills.

On the design of a flexible stimulator for animal studies in auditory prostheses

The present paper describes the design of two stimulators (bench-top and portable) which can be used for animal studies in cochlear implants. The bench-top stimulator is controlled by a high-speed digital output board manufactured by National Instruments and is electrically isolated. The portable stimulator is controlled by a personal digital assistant (PDA) and is based on a custom interface board that communicates with the signal processor in the PDA through the secure digital IO (SDIO) slot. Both stimulators can provide 8 charge-balanced, bipolar channels of pulsatile and analog-like electrical stimulation, delivered simultaneously, interleaved or using a combination of both modes. Flexibility is provided into the construction of arbitrary, but charge-balanced, pulse shapes, which can be either symmetric or asymmetric.

Evaluation of a glottal ARMA modeling scheme

In many speech analysis/synthesis schemes, the source of excitation for voiced speech is a train of impulses. Although this is a mathematically sound scheme, there is no physiological basis for it. The quality of speech that has been attained due to the introduction of a dynamically varying source, e.g., a parametric source mode, multipulse excitation, etc., has been found to be better than that produced using impulse excitation. In this paper, a pitch synchronous glottal ARMA analysis/synthesis scheme is proposed. A parametric voice source model is discussed. The voice source and vocal tract parameters are simultaneously estimated. The AR and MA orders of the vocal tract model are variable and their values are decided every period depending on whether a resynthesis efficiency threshold is crossed during the analysis/synthesis procedure. This scheme is compared with two other schemes, viz., (1) closed phase LPC analysis/synthesis and (2) robust LPC analysis/synthesis. The superiority of the proposed schem...

Effect of envelope lowpass filtering on consonant and melody recognition

Recent work [Smith et al., Nature 416, 87–90 (2002)] has shown that the speech envelope contains fine temporal information which is used in pitch perception and spatial localization. This study was performed on normal hearing subjects. In this paper, we investigated the effect of lowpass filtering of the envelope on consonant and melody recognition in subjects using the Clarion cochlear implant. The subjects were originally fitted with the simultaneous analog stimulation (SAS) speech processing strategy, a strategy known to provide fine time‐envelope information. The consonants and instrumental music were bandpass filtered into seven channels and the envelope of each channel was lowpass filtered with cutoff frequencies ranging between 100 and 1200 Hz. Initial results on the consonant recognition task showed that some subjects performed equally well for all envelope cutoff frequencies. On the melody recognition task, some subjects performed best at a particular envelope cutoff frequency. Results for the fu...

A nonlinear finite‐element model of the vocal fold.

A large‐displacement large‐strain 3‐D finite‐element model of the vocal fold was developed. The structure is discretized into 720 elements with 3003 displacement and 720 pressure degrees of freedom. The model incorporates material and geometric nonlinearities. For the constitutive law, the Mooney–Rivlin rubber material formulation for an anisotropic hyperelastic material is used. Average incompressibility constraints are introduced by adding a hydrostatic pressure work term (Lagrange multiplier) to the strain energy density function. This term is a function of the bulk modulus which has the numerical equivalence of the penalty parameter. The nodal displacements and pressure are solved for independently, using a mixed displacement/pressure formulation with 8 displacement nodes (trilinear/hexahedron) and a constant (uniform) pressure term per element. Static condensation of the discontinuous pressure variable at the element level keeps the half‐bandwidth of the stiffness matrix the same as for the displacem...

Efficient Methods for Simulating a Moving Talker in a Rectangular Room

In this paper, we describe two methods for efficiently simulating the response of a microphone to a moving talker in a rectangular room. Both methods are based on an extension of the image method to moving sources. In the first method, the microphone output signal is obtained by performing a time-domain filtering operation on the original speech signal, while in the second method, a timefrequency representation of this filtering operation is used. In each case, computational load and memory requirements are considerably reduced by taking advantage of the fact that the talker velocity is much smaller than the speed of sound.

On the use of a split‐beam array for tracking a moving talker.

Microphone arrays have been used in an audio‐teleconferencing environment to pick up the speech signal from a known direction in the presence of noise and reverberation. In current algorithms, the direction of interest is obtained by measuring the output of a beamformer at a finite set of look directions and comparing it to a threshold. Unless the set of look directions is made sufficiently large, this method is not well suited to track a moving talker. In this paper a split‐beam array configuration is used for the tracking problem. The time delay between the two halves of the array is obtained by a generalized cross‐correlation method. A one‐step Kalman predictor is then used to predict the delay for the next frame. This value is used to steer the beamformer. The system has been tested on computer‐simulated data which modeled a talker moving along a linear trajectory in a reverberant room. Results indicate that this method can provide reliable estimates of the talker bearing angle in highly reverberant e...

A study of the variation of glottal pulse shape with fundamental frequency by inverse filtering of the speech wave

The acoustics of speech production is based on the concept of a source and a filter function. In current models the source of voiced sounds is represented by a quasiperiodic succession of pulses of air emitted through the glottis, as the vocal cords open and close, and the filter function is assumed to be linear and short‐time invariant. The source term or the acoustic excitation produced by the larynx is subject to variation in speech produced by a natural speaker. The analysis described here forms part of an investigation into variation of the shape of the glottal volume velocity waveform as a function of fundamental frequency. The deconvolution of the speech wave, i.e., separation of the glottal pulse signal from the vocal tract impulse response, is done by inverse filtering the speech signal using a filter derived in the closed phase of the glottal cycle. Linear predictive covariance analysis is used in estimating the inverse filter coefficients. A linear relationship between the glottal parameters, o...