摘要:

Goldberg [J. Acoust. Soc. Am.59, S97(A) (1976)] has shown that uniform techniques for segmentation and labeling can provide the initial signal‐to‐symbol transformation for speech recognition systems with reasonable accuracy and efficiency. Furthermore, the choice of parametric representation was not found to be critical for most commonly accepted representations. However, for efficiency, the computationally simplest techniques should be used to segment the utterance before more accurate (and expensive) spectral representations are used for labeling [R. Reddy, J. Acoust. Soc. Am.42, 329–47 (1967)]. To provide an initial symbolic input for both the Harpy and Hearsay‐II systems, an hierarchical, feature‐extraction based segmenter, using the ZAPDASH parameters, has been developed. After segmentation, labeling is done by a modified LPC minimum distance [F. Itakura, IEEE Trans. ASSP‐23, 67–72 (1975)]. Labeling proceeds by comparing the midpoint of each segment with stored templates (acquired by an iterative learning process from speaker‐specific training corpus) and adjusted with weights according to features obtained from the segmenter. The use of the highly efficient segmentation procedures and parameters provides approximately a factor of 5 speedup over uniform techniques which were previously used with both Harpy and Hearsay‐II [Research supported by the Defense Advanced Projects Agency.]

ISSN:0001-4966    

DOI:10.1121/1.2003140

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

A connected digit recognition system that uses a statistical decision approach based on an expanded form of the principle of minimum residual error has been developed. The expanded distance measure includes the effects of analysis estimation error, the effects of coarticulation, and the effects of speaker variability. The recognition system has been tested on six speakers in a speaker dependent mode with recognition accuracies near 100%. It has also been tested with ten new speakers in a speaker independent mode, with a digit recognition accuracy exceeding 95%.

ISSN:0001-4966    

DOI:10.1121/1.2003179

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

The purpose of this paper is to investigate the applicability of long‐term feature averaging as an eventual means for performing text independent voice authentication (speaker verification). Based upon a set of long‐term feature vectors, a principal component analysis is performed to obtain a normalized reference coordinate system for each speaker. Features extracted from the test speaker are transformed to this coordinate system and then the Euclidean distance is measured. It is shown thatunder a weak assumption of Gaussian statistics, the threshold necessary to attain a given probability of correct acceptance as a function of the number of dimensions or features can be theoretically calculated. Results of several preliminary experiments are presented to illustrate the technique.

ISSN:0001-4966    

DOI:10.1121/1.2003182

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

Acoustic signals. structure borne, have long been used to detect the integrity of various machinery. This presentation discusses the state of the art of using the ultrasound (high frequencies beyond the audible range) of machine elements and structures to detect failure or malfunction conditions. The paper will describe a recently developed technique, the High‐Frequency Resonance Technique (HFRT), its principles, and results of applications in detecting rotating machine element failure especially in bearing inspection, as well as its potential for extensive machinery condition monitoring. A brief discussion of the use of fiber optics in the monitoring of vibrations and detection of flaws in rolling element bearing will be presented. This relatively new technique shows promise for fault detection in the areas of inspection and vibration measurement in general. The paper will also discuss the use of high frequency signal detection of pulses generated in structures failing under dynamic loads and the potential of the method for nondestructive testing of various structures in important applications.

ISSN:0001-4966    

DOI:10.1121/1.2003184

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

It has been known for many years that sound stimuli can modify the electrical activity of the brain. The possibility of developing useful clinical tests of hearing based upon this fact has also long been recognized. Efforts to create such tests, however, have until recently yielded relatively unimpressive results. The developments of the past five years are another matter, and this session will review them for interested psychologists and physiologists, and for specialists in instrumentation related to audiometry. New information on brain events time locked to stimulus presentation—the so‐called evoked responses—will be emphasized, along with the recording methods presently used to visualize them. Speakers will describe the way these evoked potentials are being used in the testing of hard‐of‐hearing patients. By emphasizing certain limitations of the methods currently employed they will also point to the future developments by which electric response audiometry may become an even more powerful clinical tool.

ISSN:0001-4966    

DOI:10.1121/1.2003188

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

Different waveforms of auditory evoked potentials were found when recording from locations in the cochlea, middle ear, and vertex of the skull in the guinea pig. The first negative potentialN1maintained its latency and waveform from a variety of sites, However, theN2potential changed its peak latency and waveform considerably with location. At certain electrode sites, double‐peakedN2waveforms were found. The data suggest that theN2potential is a complex potential reflecting activity from more than one neural population. Thus,N2is composed of repetitive firing of cochlear afferents [Özdamar, Ph.D. dissertation, Northwestern University (1976)] and neurons in the cochlear nucleus or brainstem. The data do not suggest the apical portion of the cochlea as an origin ofN2. TheN2potential was found to be most prominent from the vertex; andN1,N2, andN3appeared at nearly the same amplitudes from certain positions. The potentials maintained their characteristic wave shapes over a range of intensities. [Supported by grants from the NINCDS.]

ISSN:0001-4966    

DOI:10.1121/1.2003190

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

Transducer arrays for many underwater sound applications have customarily been designed for equal sidelobe levels. This is, perhaps, as much due to the convenience of the analytical Dolph—Chebyshev technique as to any specific requirement. Using a simple adaptation of a previously reported numerical technique for symmetrical arrays [G. L. Wilson, J. Acoust. Soc. Am.59, 195 (1976)], designs can be obtained whose directional response has a taper on the envelope enclosing the side lobes. The method is equally applicable to sum and to difference patterns. Examples are given. [This work was sponsored by the Naval Sea Systems Command, Code SEA‐034.]

ISSN:0001-4966    

DOI:10.1121/1.2003198

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

The problem of maximizing array gain (AG) for situations where ambient noise is the dominant interference has already been extensively investigated, These investigations have resulted in mathematical solutions having significant promise for various practical applications. A closely related and more general problem arises when interference is dominated by reverberation; in this situation the gain against reverberation (the reverberation gain, RG) must be maximized. The same general mathematical approach can be applied to both problems in that the omnidirectional ambient noise limited condition can be considered a specific case of the more general directional ambient noise limited condition and the reverberation limited condition. Much of the existing theory has been directed at the specific cases thus leaving unfinished a comprehensive treatment of the general case. In this presentation, the complete RG‐maximization problem will be addressed; a mathematical procedure to achieve the solution will be outlined; this procedure is developed with sufficient generality so that the maximization of AG is automatically included. An application of this technique is carried out for a three‐dimensional array which demonstrates the validity and the effectiveness of this technique. [Work supported by NUSC.]

ISSN:0001-4966    

DOI:10.1121/1.2003208

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

Although with the advent of the fast Fourier transform and digital filtering the constraints of available analog filter specifications are no longer really pertinent, the tendency to use the ANSI standard frequency bands for reference and comparison is widespread, particularly the so‐called13‐octavebands. However, in problems involving complex spectra with shifting narrow‐band components, it becomes obvious that the13‐octaveapproach is not adequate and may give misleading results, particularly at low frequencies, where the bandwidth is narrow. Similarly, because of their conceptual nature and also the way in which the bands are specified, consistent subdivision into narrow proportional bands is not possible, and such subdivision is now desirable, if not essential. For this reason, a return to the decimal system, in name as well as in fact, is strongly recommended. The problems encountered with standard “13‐octave” bands are discussed in detail, in the context of the band specifications, and alternative methods of band specification, which avoid the problems entirely, are described, including a new look at the concept of “spectrum level.” [Work supported by NUSC.]

ISSN:0001-4966    

DOI:10.1121/1.2003209

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP

摘要:

This paper describes measurements of noise from cavitating flow over a hydrofoil. Cavitation was generated on a hydrofoil in the presence of a separated laminar boundary layer on the one hand and of a fully turbulent attached boundary layer on the other. The turbulent boundary layer was formed down‐stream of a trip which was positioned near the leading edge. The noise is shown to depend on the type of cavitation produced. Dimensionless spectral densities of the sound are shown for each type of flow. For the cavitation associated with turbulent boundary layer development, the dimensionless spectrum is interpreted in terms of the observed single‐bubble motions in the experiment.

ISSN:0001-4966    

DOI:10.1121/1.2003218

出版商:Acoustical Society of America    

年代:1976

数据来源: AIP