T.J Moir - Academia.edu (original) (raw)

Papers by T.J Moir

5th European Conference on Speech Communication and Technology (Eurospeech 1997)

This paper introduces a word boundary detection algorithm that works in a variety of noise condit... more This paper introduces a word boundary detection algorithm that works in a variety of noise conditions including what is commonly called the 'cocktail party' situation. The algorithm uses the direction of the signal as the main criterion for differentiating between desired-speech and background noise. To determine the signal direction the algorithm calculates estimates of the time delay between signals received at two microphones. These time delay estimates together with estimates of the coherence function and signal energy are used to locate word boundaries. The algorithm was tested using speech embedded in different types and levels of noise including car noise, factory noise, babble noise, and competing talkers. The test results showed that the algorithm performs very well under adverse conditions and with SNR down to-14.5dB.

A combined three-microphone voice activity detector (VAD) and noise-canceling system is studied t... more A combined three-microphone voice activity detector (VAD) and noise-canceling system is studied to enhance speech recognition in an automobile environment. A previous experiment clearly shows the ability of the composite system to cancel a single noise source outside of a defined zone. This paper investigates the performance of the composite system when there are frequently moving noise sources (noise sources are coming from different locations but are not always presented at the same time) e.g. there is other passenger speech or speech from a radio when a desired speech is presented. To work in a frequently moving noise sources environment, whilst a three-microphone voice activity detector (VAD) detects voice from a "VAD valid zone", the 3-microphone noise canceller uses a "noise canceller valid zone" defined in freespace around the users head. Therefore, a desired voice should be in the intersection of the noise canceller valid zone and VAD valid zone. Thus all...

A kepstrum approach to filtering, smoothing and prediction with application to speech enhancement

Abstract: The method of steepest-descent is re-visited in continuous time. It is shown that the c... more Abstract: The method of steepest-descent is re-visited in continuous time. It is shown that the continuous time version is a vector differential equation the solution of which is found by integration. Since numerical integration has many forms, we show an alternative to the conventional solution by using a Trapezoidal integration solution. This in turn gives a slightly modified least-mean squares (LMS) algorithm.

2005 International Conference on Intelligent Sensors, Sensor Networks and Information Processing, 2005

This paper addresses issues in improving hands-free speech recognition performance in car environ... more This paper addresses issues in improving hands-free speech recognition performance in car environments. A three-microphone array has been used to form a beamformer with leastmean squares (LMS) to improve Signal to Noise Ratio (SNR). A three-microphone array has been paralleled to a Voice Activity Detection (VAD). The VAD uses time-delay estimation together with magnitude-squared coherence (MSC).

Thispaperpresents a novelin-car designto recognize thedriver's voiceandcontrol in-car devices... more Thispaperpresents a novelin-car designto recognize thedriver's voiceandcontrol in-car devices. Ina Microphone 1 LabVIEWprogram non-stationary noise carenvironment, aspeechrecognition kit F

2016 3rd International Conference on Signal Processing and Integrated Networks (SPIN), 2016

A gain matrix is used in order to study the optimal step-size required for the least-mean squares... more A gain matrix is used in order to study the optimal step-size required for the least-mean squares (LMS) algorithm. The case of additive measurement noise is studied and it is proven that the optimal gain matrix based LMS algorithm developed is identical to a Kalman filter. The key importance of this proof of equivalence is that it is shown that for correlated driving noise, the LMS algorithm is always sub-optimal. For white driving noise the optimal step-size is scalar and time-varying.

International Journal of Systems Science, 2017

ABSTRACT The Wiener–Kolmogorov theory of filtering has been with us since the first half of the t... more ABSTRACT The Wiener–Kolmogorov theory of filtering has been with us since the first half of the twentieth century. A later matrix-based approach which was more general was derived with the steady-state Kalman filter. This approach uses a novel method of representing causal and uncausal systems in the form of convolution matrices and leads to a Wiener solution which is much easier to calculate than either the Kalman or Wiener approaches. For coloured additive noise, it avoids the use of Diophantine equations. The key idea missing in previous work is the close link between polynomials and Toeplitz matrices which are lower triangular in form. There is already a reasonably sized literature in the mathematics field on such matrices and so the area is ripe for exploration. Although the method does not offer a different or better solution, it shows a completely new way of defining linear time-invariant (LTI) systems which is neither transfer-function nor state-space-based. This is achieved by exploiting the connection between polynomials and Toeplitz matrices. The application here is the Wiener filter but there could well be many more as this is a generic approach.

IEE Proceedings D Control Theory and Applications, 1991

Abstract A state-space self-tuning LQG regulator is obtained. The algorithm differs from other st... more Abstract A state-space self-tuning LQG regulator is obtained. The algorithm differs from other state-space approaches since the controller and estimator are found by polynomial equations rather than matrix manipulation. The approach arises naturally from the Wiener theory of control, rather than the steady-state Kalman theory of other work. An example is given of the control of an open-loop, unstable and nonminimum phase system

IFAC Proceedings Volumes, 1983

The weighted minimum variance controller is employed in a selftuning control system for plants wh... more The weighted minimum variance controller is employed in a selftuning control system for plants which can be multivariable and non-square. Both explicit and implicit self-tuning strategies are discussed.

Signal, Image and Video Processing, 2015

A novel approach to finite-impulse response system identification is given. The method is formula... more A novel approach to finite-impulse response system identification is given. The method is formulated differently from ordinary least mean squares (LMS) or block LMS, which are traditional approaches and has a significant advantage of speed improvement when the correlation matrix has a large condition number. Unlike other approaches which are numerically complex, this method has a similar computational burden as LMS and gives the same optimal solution after convergence. It avoids transformation matrices by writing the system description in terms of a convolution matrix which has a special lower-triangular format. In this way the correlation matrix is different from conventional least-squares approaches and maintains a modest condition number as the correlation matrix in ordinary least-squares climbs high. The method has as wide a range of applications as ordinary LMS-based solutions but can also work on deterministic inputs.

Laser Radar Technology and Applications III, 1998

In this paper, the demodulation characteristics of an amplitude locked loop (ALL) integrated into... more In this paper, the demodulation characteristics of an amplitude locked loop (ALL) integrated into a laser sensing system are presented. The laser sensing system used here is a free space laser Doppler vibrometer which could be used for detecting the frequency of vibration of an object. The sensitivity of such a device is impaired by interference arising from spurious scattering along the propagation path. Mathematical and experimental investigations concluded that this interference pattern is identical to cochannel interference encountered in conventional radio-wave frequency modulation. By incorporating an ALL with a phase locked loop in the demodulator of the optical system, a dramatic improvement over 10 dB Sinad (measurement done with no psophometer) was found over existing technology.

Optical Inspection and Micromeasurements, 1996

A novel application of the incorporation of an amplitude- locked loop (ALL) into a coherent optic... more A novel application of the incorporation of an amplitude- locked loop (ALL) into a coherent optical sensing system is presented. The ALL is the mathematical dual of the phase- locked loop and is a unique circuit which operates on amplitude rather than phase. Current work on the ALL has concentrated on developing a simple free space Doppler velocimetry system whereby an object's velocity may be identified by analyzing the frequency content of coherent light reflected from the object. It is the purpose of this paper to highlight the potential advantages of using an ALL in the detection circuitry of a coherent optical sensing system in order to improve the detection sensitivity of such a system.

The two input Griffiths-Jim acoustic beamformer is analysed in the frequency domain using a Wiene... more The two input Griffiths-Jim acoustic beamformer is analysed in the frequency domain using a Wiener type formulation. Unlike previous solutions the approach here is to look at the problem of non-minimum phase acoustic transfer functions which are encountered in many real filtering problems. The polynomial transfer function approach gives an elegant way of obtaining the frequency response of the beamformer and gives new insight to the problem in general.

International Journal of Control, 1986

The Fourier transform of the logarithm of spectral density is a useful tool for spectral analysis... more The Fourier transform of the logarithm of spectral density is a useful tool for spectral analysis of random signals which are highly resonant. This is because the logarithm compresses the large peaks of the spectrum and a resulting power series expansion (kepstrum) can be truncated at a suitable length to suppress the higher frequencies. This paper utilizes the FFT in a similar form in order to obtain spectral smoothing. Several examples show the advantages of the method including an analysis on the pitch and roll data of a container ...

International Journal of Computer Applications in Technology, 2010

This paper explores the problems of speech recognition in a (sometimes) noisy environment. An ada... more This paper explores the problems of speech recognition in a (sometimes) noisy environment. An adaptive acoustic beamformer is proposed based on the Griffiths-Jim method and a 'hot-spot' where speech can be received within a geometric-defined boundary and rejected outside of it will be shown to give a certain amount of noise immunity and improve the signal-tonoise ratio for the second stage, which is the speech recognition engine. The recognition engine used has a limited vocabulary which gives rise to an excellent hit-rate and less training than unlimited vocabulary. The technology here has improved vastly within the last decade and it will be shown that by using a head and shoulders avatar that is both photo-realistic and with appealing personality, the experience of a speech interface is vastly enhanced. The paper will explore these technologies and investigate the convergence of many of them in the current Massey smart-office.

International Journal of Computer Applications in Technology, 2010

This paper presents a novel in-car design to recognise the driver's voice and to control in-car d... more This paper presents a novel in-car design to recognise the driver's voice and to control in-car devices. In a non-stationary noise car environment, a speech recognition kit SR-07 with speech recognition chip HM2007 has less than 10% successful hit-rate. With an adaptive Wiener filter, the successful hit-rate has improved to as much as 75%.

International Journal of Adaptive Control and Signal Processing, 2012

SUMMARY A new structured approach to adaptive noise cancelling of non-stationary stochastic signa... more SUMMARY A new structured approach to adaptive noise cancelling of non-stationary stochastic signals is given. The divide and conquer method subdivides the problem into a number of sources with a power of 2 and subdivides the solution to smaller problems in a similar fashion to that of an FFT. Hence, with the number of sources 2 p, where p is a positive integer, the problem is successively reduced to simpler solutions at each stage of order 2 p− 1. In its basic form, the method uses only multivariable least mean squares (MLMS), ...

International Journal of Adaptive Control and Signal Processing, 2009

In real‐time environments a speech recognition system in a car has to receive the driver's vo... more In real‐time environments a speech recognition system in a car has to receive the driver's voice only while suppressing the background noise. This paper presents a hybrid real‐time adaptive filter that operates within a geometrical zone defined around the head of the desired speaker. Any sound outside of this zone is considered to be noise and is suppressed. As this defined geometrical zone is small, it is assumed that only driver's speech is incoming from this zone. The technique uses three microphones to define a geometric‐based voice–activity detector (VAD) to cancel the unwanted speech coming from outside of the zone. However, when unwanted speech and desired speech are incoming at the same time, the VAD fails to identify the unwanted speech or desired speech. In such a situation an adaptive Wiener filter is switched on for noise reduction. In the case of sole unwanted speech incoming from outside of a desired zone, this speech is muted at the output of the hybrid noise ...

Electronics Letters, 1992

5th European Conference on Speech Communication and Technology (Eurospeech 1997)

This paper introduces a word boundary detection algorithm that works in a variety of noise condit... more This paper introduces a word boundary detection algorithm that works in a variety of noise conditions including what is commonly called the 'cocktail party' situation. The algorithm uses the direction of the signal as the main criterion for differentiating between desired-speech and background noise. To determine the signal direction the algorithm calculates estimates of the time delay between signals received at two microphones. These time delay estimates together with estimates of the coherence function and signal energy are used to locate word boundaries. The algorithm was tested using speech embedded in different types and levels of noise including car noise, factory noise, babble noise, and competing talkers. The test results showed that the algorithm performs very well under adverse conditions and with SNR down to-14.5dB.

A combined three-microphone voice activity detector (VAD) and noise-canceling system is studied t... more A combined three-microphone voice activity detector (VAD) and noise-canceling system is studied to enhance speech recognition in an automobile environment. A previous experiment clearly shows the ability of the composite system to cancel a single noise source outside of a defined zone. This paper investigates the performance of the composite system when there are frequently moving noise sources (noise sources are coming from different locations but are not always presented at the same time) e.g. there is other passenger speech or speech from a radio when a desired speech is presented. To work in a frequently moving noise sources environment, whilst a three-microphone voice activity detector (VAD) detects voice from a "VAD valid zone", the 3-microphone noise canceller uses a "noise canceller valid zone" defined in freespace around the users head. Therefore, a desired voice should be in the intersection of the noise canceller valid zone and VAD valid zone. Thus all...

A kepstrum approach to filtering, smoothing and prediction with application to speech enhancement

Abstract: The method of steepest-descent is re-visited in continuous time. It is shown that the c... more Abstract: The method of steepest-descent is re-visited in continuous time. It is shown that the continuous time version is a vector differential equation the solution of which is found by integration. Since numerical integration has many forms, we show an alternative to the conventional solution by using a Trapezoidal integration solution. This in turn gives a slightly modified least-mean squares (LMS) algorithm.

2005 International Conference on Intelligent Sensors, Sensor Networks and Information Processing, 2005

This paper addresses issues in improving hands-free speech recognition performance in car environ... more This paper addresses issues in improving hands-free speech recognition performance in car environments. A three-microphone array has been used to form a beamformer with leastmean squares (LMS) to improve Signal to Noise Ratio (SNR). A three-microphone array has been paralleled to a Voice Activity Detection (VAD). The VAD uses time-delay estimation together with magnitude-squared coherence (MSC).

Thispaperpresents a novelin-car designto recognize thedriver's voiceandcontrol in-car devices... more Thispaperpresents a novelin-car designto recognize thedriver's voiceandcontrol in-car devices. Ina Microphone 1 LabVIEWprogram non-stationary noise carenvironment, aspeechrecognition kit F

2016 3rd International Conference on Signal Processing and Integrated Networks (SPIN), 2016

A gain matrix is used in order to study the optimal step-size required for the least-mean squares... more A gain matrix is used in order to study the optimal step-size required for the least-mean squares (LMS) algorithm. The case of additive measurement noise is studied and it is proven that the optimal gain matrix based LMS algorithm developed is identical to a Kalman filter. The key importance of this proof of equivalence is that it is shown that for correlated driving noise, the LMS algorithm is always sub-optimal. For white driving noise the optimal step-size is scalar and time-varying.

International Journal of Systems Science, 2017

ABSTRACT The Wiener–Kolmogorov theory of filtering has been with us since the first half of the t... more ABSTRACT The Wiener–Kolmogorov theory of filtering has been with us since the first half of the twentieth century. A later matrix-based approach which was more general was derived with the steady-state Kalman filter. This approach uses a novel method of representing causal and uncausal systems in the form of convolution matrices and leads to a Wiener solution which is much easier to calculate than either the Kalman or Wiener approaches. For coloured additive noise, it avoids the use of Diophantine equations. The key idea missing in previous work is the close link between polynomials and Toeplitz matrices which are lower triangular in form. There is already a reasonably sized literature in the mathematics field on such matrices and so the area is ripe for exploration. Although the method does not offer a different or better solution, it shows a completely new way of defining linear time-invariant (LTI) systems which is neither transfer-function nor state-space-based. This is achieved by exploiting the connection between polynomials and Toeplitz matrices. The application here is the Wiener filter but there could well be many more as this is a generic approach.

IEE Proceedings D Control Theory and Applications, 1991

Abstract A state-space self-tuning LQG regulator is obtained. The algorithm differs from other st... more Abstract A state-space self-tuning LQG regulator is obtained. The algorithm differs from other state-space approaches since the controller and estimator are found by polynomial equations rather than matrix manipulation. The approach arises naturally from the Wiener theory of control, rather than the steady-state Kalman theory of other work. An example is given of the control of an open-loop, unstable and nonminimum phase system

IFAC Proceedings Volumes, 1983

The weighted minimum variance controller is employed in a selftuning control system for plants wh... more The weighted minimum variance controller is employed in a selftuning control system for plants which can be multivariable and non-square. Both explicit and implicit self-tuning strategies are discussed.

Signal, Image and Video Processing, 2015

A novel approach to finite-impulse response system identification is given. The method is formula... more A novel approach to finite-impulse response system identification is given. The method is formulated differently from ordinary least mean squares (LMS) or block LMS, which are traditional approaches and has a significant advantage of speed improvement when the correlation matrix has a large condition number. Unlike other approaches which are numerically complex, this method has a similar computational burden as LMS and gives the same optimal solution after convergence. It avoids transformation matrices by writing the system description in terms of a convolution matrix which has a special lower-triangular format. In this way the correlation matrix is different from conventional least-squares approaches and maintains a modest condition number as the correlation matrix in ordinary least-squares climbs high. The method has as wide a range of applications as ordinary LMS-based solutions but can also work on deterministic inputs.

Laser Radar Technology and Applications III, 1998

In this paper, the demodulation characteristics of an amplitude locked loop (ALL) integrated into... more In this paper, the demodulation characteristics of an amplitude locked loop (ALL) integrated into a laser sensing system are presented. The laser sensing system used here is a free space laser Doppler vibrometer which could be used for detecting the frequency of vibration of an object. The sensitivity of such a device is impaired by interference arising from spurious scattering along the propagation path. Mathematical and experimental investigations concluded that this interference pattern is identical to cochannel interference encountered in conventional radio-wave frequency modulation. By incorporating an ALL with a phase locked loop in the demodulator of the optical system, a dramatic improvement over 10 dB Sinad (measurement done with no psophometer) was found over existing technology.

Optical Inspection and Micromeasurements, 1996

A novel application of the incorporation of an amplitude- locked loop (ALL) into a coherent optic... more A novel application of the incorporation of an amplitude- locked loop (ALL) into a coherent optical sensing system is presented. The ALL is the mathematical dual of the phase- locked loop and is a unique circuit which operates on amplitude rather than phase. Current work on the ALL has concentrated on developing a simple free space Doppler velocimetry system whereby an object's velocity may be identified by analyzing the frequency content of coherent light reflected from the object. It is the purpose of this paper to highlight the potential advantages of using an ALL in the detection circuitry of a coherent optical sensing system in order to improve the detection sensitivity of such a system.

The two input Griffiths-Jim acoustic beamformer is analysed in the frequency domain using a Wiene... more The two input Griffiths-Jim acoustic beamformer is analysed in the frequency domain using a Wiener type formulation. Unlike previous solutions the approach here is to look at the problem of non-minimum phase acoustic transfer functions which are encountered in many real filtering problems. The polynomial transfer function approach gives an elegant way of obtaining the frequency response of the beamformer and gives new insight to the problem in general.

International Journal of Control, 1986

The Fourier transform of the logarithm of spectral density is a useful tool for spectral analysis... more The Fourier transform of the logarithm of spectral density is a useful tool for spectral analysis of random signals which are highly resonant. This is because the logarithm compresses the large peaks of the spectrum and a resulting power series expansion (kepstrum) can be truncated at a suitable length to suppress the higher frequencies. This paper utilizes the FFT in a similar form in order to obtain spectral smoothing. Several examples show the advantages of the method including an analysis on the pitch and roll data of a container ...

International Journal of Computer Applications in Technology, 2010

This paper explores the problems of speech recognition in a (sometimes) noisy environment. An ada... more This paper explores the problems of speech recognition in a (sometimes) noisy environment. An adaptive acoustic beamformer is proposed based on the Griffiths-Jim method and a 'hot-spot' where speech can be received within a geometric-defined boundary and rejected outside of it will be shown to give a certain amount of noise immunity and improve the signal-tonoise ratio for the second stage, which is the speech recognition engine. The recognition engine used has a limited vocabulary which gives rise to an excellent hit-rate and less training than unlimited vocabulary. The technology here has improved vastly within the last decade and it will be shown that by using a head and shoulders avatar that is both photo-realistic and with appealing personality, the experience of a speech interface is vastly enhanced. The paper will explore these technologies and investigate the convergence of many of them in the current Massey smart-office.

International Journal of Computer Applications in Technology, 2010

This paper presents a novel in-car design to recognise the driver's voice and to control in-car d... more This paper presents a novel in-car design to recognise the driver's voice and to control in-car devices. In a non-stationary noise car environment, a speech recognition kit SR-07 with speech recognition chip HM2007 has less than 10% successful hit-rate. With an adaptive Wiener filter, the successful hit-rate has improved to as much as 75%.

International Journal of Adaptive Control and Signal Processing, 2012

SUMMARY A new structured approach to adaptive noise cancelling of non-stationary stochastic signa... more SUMMARY A new structured approach to adaptive noise cancelling of non-stationary stochastic signals is given. The divide and conquer method subdivides the problem into a number of sources with a power of 2 and subdivides the solution to smaller problems in a similar fashion to that of an FFT. Hence, with the number of sources 2 p, where p is a positive integer, the problem is successively reduced to simpler solutions at each stage of order 2 p− 1. In its basic form, the method uses only multivariable least mean squares (MLMS), ...

International Journal of Adaptive Control and Signal Processing, 2009

In real‐time environments a speech recognition system in a car has to receive the driver's vo... more In real‐time environments a speech recognition system in a car has to receive the driver's voice only while suppressing the background noise. This paper presents a hybrid real‐time adaptive filter that operates within a geometrical zone defined around the head of the desired speaker. Any sound outside of this zone is considered to be noise and is suppressed. As this defined geometrical zone is small, it is assumed that only driver's speech is incoming from this zone. The technique uses three microphones to define a geometric‐based voice–activity detector (VAD) to cancel the unwanted speech coming from outside of the zone. However, when unwanted speech and desired speech are incoming at the same time, the VAD fails to identify the unwanted speech or desired speech. In such a situation an adaptive Wiener filter is switched on for noise reduction. In the case of sole unwanted speech incoming from outside of a desired zone, this speech is muted at the output of the hybrid noise ...

Electronics Letters, 1992