Publications
Robust speech recognition using hidden Markov models: overview of a research program
February 26, 1990
Technical Report
Published in:
MIT Lincoln Laboratory Report TR-875
Topic:
R&D area:
Summary
This report presents an overview of a program of speech recognition research which was initiated in 1985 with the major goal of developing techniques for robust high performance speech recognition under the stress and noise conditions typical of a military aircraft cockpit. The work on recognition in stress and noise during 1985 and 1986 produced a robust Hidden Markov Model (HMM) isolated-word recognition (IWR) system with 99 percent speaker-dependent accuracy for several difficult stress/noise data bases, and very high performance for normal speech. Robustness techniques which were developed and applied include multi-style training, robust estimation of parameter variances, perceptually-motivated stress-tolerant distance measures, use of time-differential speech parameters, and discriminant analysis. These techniques and others produced more than an order-of-magnitude reduction in isolated-word recognition error rate relative to a baseline HMM system. An important feature of the Lincoln HMM system has been the use of continuous-observation HMM techniques, which provide a good basis for the development of the robustness techniques, and avoid the need for a vector quantizer at the input to the HMM system. Beginning in 1987, the robust HMM system has been extended to continuous speech recognition for both speaker-dependent and speaker-independent tasks. The robust HMM continuous speech recognizer was integrated in real-time with a stressing simulated flight task, which was judged to be very realistic by a number of military pilots. Phrase recognition accuracy on the limited-task-domain (28-word vocabulary) flight task is better than 99.9 percent. Recently, the robust HMM system has been extended to large-vocabulary continuous speech recognition, and has yielded excellent performance in both speaker-dependent and speaker-independent recognition on the DARPA 1000-word vocabulary resource management data base. Current efforts include further improvements to the HMM system, techniques for the integration of speech recognition with natural language processing, and research on integration of neural network techniques with HMM.
Summary
This report presents an overview of a program of speech recognition research which was initiated in 1985 with the major goal of developing techniques for robust high performance speech recognition under the stress and noise conditions typical of a military aircraft cockpit. The work on recognition in stress and noise...
READ MORE
Low-sidelobe phased array antenna characteristics using the planar near-field scanning technique: theory and experiment
February 14, 1990
Technical Report
Published in:
MIT Lincoln Laboratory Report TR-870
Topic:
R&D group:
Summary
Characteristics of a low-sidelobe phased array antenna are investigated using the technique of planar near-field scanning. The theory associated with the planar near-field scanning technique, with and without probe compensation, is reviewed and an application of the theory is made. The design of an experimental low-sidelobe phased array antenna consisting of monopole elements which are corporate-fed using high precision transmit/receive modules is described. Accurate array radiation patterns are obtained both theoretically and experimentally using centerline scanning at less than one wavelength distance from the antenna. The effects of the antenna probe on the array near-field pattern, plane-wave spectrum, and far-field pattern are demonstrated theoretically using a method of moments numerical simulation. Comparisons of the array theoretical near-zone electric field and array received voltage due to a V-dipole near-field transmitting probe are made. It is shown that a V-dipole theoretical probe antenna can accurately model a practical near-field measurement probe consisting of an open-ended rectangular waveguide surrounded with anechoic material.
Summary
Characteristics of a low-sidelobe phased array antenna are investigated using the technique of planar near-field scanning. The theory associated with the planar near-field scanning technique, with and without probe compensation, is reviewed and an application of the theory is made. The design of an experimental low-sidelobe phased array antenna consisting...
READ MORE
Use of clutter residue editing maps during the Denver 1988 Terminal Doppler Weather Radar (TDWR) tests
January 25, 1990
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-169
Summary
The Lincoln Laboratory Terminal Doppler Weather Radar (TDWR) testbed operated in Denver, CO in 1987-88. This radar is a prototype of the wind shear detection radars scheduled to be installed by the FAA to provide warnings of possibly hazardous wind shear conditions in airport terminal areas. To obtain the required coverage at low altitudes (down to 100-200 meters above ground level), the antenna beam is required to scan at or very near the earth's surface. Strong ground clutter returns at these low elevation angles present a major problem in the detection of low reflectivity wind shear signals and pose a significant challenge to the mission of these radars. To address this problem, steps along several fronts are taken to mitigate the effects of clutter contamination. These include the use of narrow pencil-beam antennas to minimize ground illumination, suppression by high-pass clutter filters, and the use of clutter residue map editing. This report deals with the latter step, and focuses on the clutter environment experienced at the testbed site during April-October 1988 and its effect on clutter residue map usage. Since the clutter environment is subject to change over time -- due either to man-made or natural causes -- the residue maps require periodic updates to reflect the changing nature of the clutter. This is particularly important for radar systems such as these which rely on automated algorithms to detect subtle patterns and features in the radar returns. To study the frequency with which residue maps required replacement in Denver, clutter measurements recorded during this period were analyzed and are presented in this report as a series of clutter residue maps. The maps are compared and the short and long term changes analyzed. It is concluded that the overall changes during this time were relatively small and gradual, and that map updates at one to two month intervals were sufficient. The generation of the residue maps is described and the importance of collecting clutter data on clear, weather-free days, without the presence of anomalous propagation conditions is addressed. This report also describes the use of median estimation in the construction of the maps as an effective method of eliminating the occasional strong returns from moving reflectors, such as aircraft and vehicles, which would otherwise distort the maps.
Summary
The Lincoln Laboratory Terminal Doppler Weather Radar (TDWR) testbed operated in Denver, CO in 1987-88. This radar is a prototype of the wind shear detection radars scheduled to be installed by the FAA to provide warnings of possibly hazardous wind shear conditions in airport terminal areas. To obtain the required...
READ MORE
Very low grazing angle backscatter from the ocean surface
January 4, 1990
Conference Paper
Published in:
IEEE Int. Radar Conf. 1990, 6-9 May 1990, pp. 181-188.
Topic:
R&D group:
Summary
A very low grazing angle (
Summary
A very low grazing angle (
READ MORE
An approach to co-channel talker interference suppression using a sinusoidal model for speech
January 1, 1990
Journal Article
Published in:
IEEE Trans. Acoust. Speech Signal Process., Vol. 38, No. 1, January 1990, pp. 56-59.
Topic:
R&D area:
Summary
This paper describes a new approach to co-channel talker interference suppression on a sinusoidal representation of speech. The technique fits a sinusoidal model to additive vocalic speech segments such that the least mean-squared error between the model and the summed waveforms is obtained. Enhancement is achieved by synthesizing a waveform from the sine waves attributed to the desired speaker. Least-squares estimation is applied to obtain sine-wave amplitudes and phases of both talkers, based on either a priori sine-wave frequencies or a priori fundamental frequency contours. When the frequencies of the two waveforms are closely spaced, the performance is significantly improved by exploiting the time evolution of the sinusoidal parameters across multiple analysis frames. The least-squared error approach is also extended, under restricted conditions, to estimate fundamental frequency contours of both speakers from the summed waveforms. The results obtained, although limited in their scope, provide evidence that the sinusoidal analysis/synthesis model with effective parameter estimation techniques offers a promising approach to the problem of co-channel talker interference suppression over a range of conditions.
Summary
This paper describes a new approach to co-channel talker interference suppression on a sinusoidal representation of speech. The technique fits a sinusoidal model to additive vocalic speech segments such that the least mean-squared error between the model and the summed waveforms is obtained. Enhancement is achieved by synthesizing a waveform...
READ MORE
Spoken language systems
January 1, 1990
Journal Article
Published in:
Annu. Rev. Comput. Sci., Vol. 4, 1990, pp. 481-501.
Topic:
R&D area:
Summary
Spoken language is the most natural and common form of human-human communication, whether face to face, over the telephone, or through various communication media such as radio and television. In contrast, human-machine interaction is currently achieved largely through keyboard strokes, pointing, or other mechanical means, using highly stylized languages. Communication, whether human-human or human-machine, suffers greatly when the two communicating agents do not "speak" the same language. The ultimate goal of work on spoken language systems is to overcome this language barrier by building systems that provide the necessary interpretive function between various languages, thus establishing spoken language as a versatile and natural communication medium between humans and machines and among humans speaking different languages.
Summary
Spoken language is the most natural and common form of human-human communication, whether face to face, over the telephone, or through various communication media such as radio and television. In contrast, human-machine interaction is currently achieved largely through keyboard strokes, pointing, or other mechanical means, using highly stylized languages. Communication...
READ MORE
A case study of the 24 August 1986, FLOWS microburst
November 28, 1989
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-162
Summary
From 1984 to 1986, Lincoln Laboratory under the sponsorship of the Federal Aviation Administration (FAA) collected wind shear measurements in the southeastern United States using a pulsed Doppler radar. The major emphasis of the measurement program and subsequent analyses is the development and testing of algorithms that will enable the Terminal Doppler Weather Radar (TDWR) to provide wind shear warnings to the aviation community by detection and tracking gust fronts and microbursts. An important phase of the program involves determining appropriate scan strategies and algorithms to detect other radar measurable features which precede or accompany the surface outflows of microbursts. The detection of features aloft such as convergence, rotation, divergence, storm cells, and descending reflectivity cores may permit advanced recognition of the wind shear while it is less than 10 m/s. In this report a microburst on 24 August 1986 in Huntsville is analyzed with single and dual-Doppler techniques to assess microburst precursors, asymmetry, and forcing mechanisms which could be used for futute algorithm development. The microburst producing storm formed within a moist adiabatic, unstable air-mass with weak wind shear at low to mid-levels of the atmosphere. Rotation, convergence, divergent tops, and a descending core were detected prior to the outflow attaining a divergence of 10 m/s. This storm is similar to other Huntsville microburst producing cells in exhibiting upper-level divergence prior to the initial microburst outflow. Previous analyses of wind shear in Denver and Oklahoma did not discuss divergent tops as a possible microburst precursor. However, its relation to storm severity and hailstorm intensity has been reported by Witt and Nelson (1984) and NEXRAD Program Office (1985). In this case-study, the 3-dimensional microburst detection algorithm provided an early declaration of the event while the radial velocity differential was less than 10 m/s.
Summary
From 1984 to 1986, Lincoln Laboratory under the sponsorship of the Federal Aviation Administration (FAA) collected wind shear measurements in the southeastern United States using a pulsed Doppler radar. The major emphasis of the measurement program and subsequent analyses is the development and testing of algorithms that will enable the...
READ MORE
Development of an automated windshear detection system using doppler weather radar
November 1, 1989
Journal Article
Published in:
Proc. IEEE, Vol. 77, No. 11, November 1989, pp. 1661-1673.
Topic:
R&D area:
R&D group:
Summary
The US Federal Aviation Administration (FAA) is developing the Terminal Doppler Weather Radar (TDWR) system to determine the location and severity of LAWS (low-altitude windshear) phenomena and other weather hazards (e.g. tornadoes and turbulence) and to provide the pertinent information to real-time air traffic control users. The FAA program for developing and evaluating the TDWR is described, with emphasis on the resolution of key technical issues such as separation of the radar return due to the low-altitude weather phenomena from that caused by various clutter sources and the automatic detection of the phenomena by means of pattern recognition applied to images depicting the weather reflectivity and Doppler shift. These technical issues have been addressed using experimental data obtained using a testbed radar in representative meteorological regimes. The system performance has been assessed using numerous experimental windshear data sets with corresponding 'truth' developed by experienced radar meteorologists from a number of organizations. It is shown that the system provides very reliable detection of strong microbursts in a variety of environments with a gust-front detection capability that supports effective planning of airport runway use.
Summary
The US Federal Aviation Administration (FAA) is developing the Terminal Doppler Weather Radar (TDWR) system to determine the location and severity of LAWS (low-altitude windshear) phenomena and other weather hazards (e.g. tornadoes and turbulence) and to provide the pertinent information to real-time air traffic control users. The FAA program for...
READ MORE
The relationship between lightning type and convective state of thunderclouds
September 30, 1989
Journal Article
Published in:
J. Geophys. Res., Vol. 94, No. D11, 30 September 1989, pp. 13,213-13,220.
Topic:
R&D area:
R&D group:
Summary
Thunderstorm case studies and earlier observations are described which illuminate the relationship between cloud vertical development and the prevalence of intracloud (IC) and cloud-to-ground (CG) lightning. A consistent temporal evolution starting with peak IC activity changing to predominant CG activity and concluding with strong outflow (microburst) suggests that ice is responsible for both the electrical (i.e., lightning) and dynamical (i.e., microburst) phenomena. The IC activity is attributed to the updraft-driven accumulation of graupel particles in the central dipole region, and the subsequent CG activity to the descent of ice particles beneath the height of the main negative charge. The subsequent descent and melting of ice particles beneath the height of the 0 degree C isotherm are associated with the acceleration of the downdraft and outflow. The IC lightning precursor can provide a valuable short-term (5-10 min) warning for microburst hazard at ground level.
Summary
Thunderstorm case studies and earlier observations are described which illuminate the relationship between cloud vertical development and the prevalence of intracloud (IC) and cloud-to-ground (CG) lightning. A consistent temporal evolution starting with peak IC activity changing to predominant CG activity and concluding with strong outflow (microburst) suggests that ice is...
READ MORE
Mode S Beacon System: a functional overview
August 29, 1989
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-150
Topic:
R&D area:
R&D group:
Summary
This document provides a functional overview of the Mode S Beacon System, a combined secondary surveillance radar (beacon) and ground-air-ground data link system capable of providing the aircraft surveillance and communications necessary to support ATC automation in future traffic environments. Mode S is capable of common-channel interoperation with the current ATC beacon system, and may be implemented over an extended transition period. Mode S will provide the surveillance and communication performance required by ATC automation, the reliable communications needed to support data link services, and the capability of operating with a terminal or enroute, radar digitizer-equipped, ATC surveillance radar. The material contained in this document serves as an introduction to the more detailed information contained in "Mode S Beacon System: Functional Description." DOT/FAA/PM-86/19, 29 August 1986.
Summary
This document provides a functional overview of the Mode S Beacon System, a combined secondary surveillance radar (beacon) and ground-air-ground data link system capable of providing the aircraft surveillance and communications necessary to support ATC automation in future traffic environments. Mode S is capable of common-channel interoperation with the current...
READ MORE