摘要:

In this talk we will be concerned primarily withacousticalmethods for estimating vocal tract shapes. These fall into two general categories: (i) methods that estimate the shape from the speech signal and (ii) those that require external measurements. Methods in the first category are based on several tacit assumptions. The validity of these assumptions is not easily tested, and the effects of departures from these assumptions are quite significant, yet difficult to quantify. Of the methods in the second category, the one based on periodically measuring the impulse response at the lips [Sondhi and Gopinath, J. Acoust. Soc. Am.49, 1867–1873 (1971)] will be shown to be potentially the most reliable. This method recovers the cross‐sectional area functions exactly for a lossless vocal tract under the single assumption of plane wave propagation. We will briefly review this method and present evidence that departures from the basic assumptions do not drastically affect the reconstructed shapes. We will also review the algorithms presently available to take account of distributed losses. Finally, we will suggest that much useful information would be gained if the impulse response measurements are made simultaneously with an x‐ray movie of the midsagittal plane.

ISSN:0001-4966    

DOI:10.1121/1.2004202

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

A two‐state model ofF0 dynamics in speech production has been developed which characterizesF0 control as a succession of alternating steady‐state and transition control states. The control states have been defined in terms of pitch period distribution parameters obtained in an earlier study [R. J. Hanson and P. F. MacNeilage, J. Acoust. Soc. Am.55, S31(P8) (1974)]. The model attempts to synthesize observations about sensorimotor integration in speech while capitalizing on differences in the perception and production of steady‐state versus transition portions of the speech signal. A preliminary version of the model has been implemented in aFORTRANprogram and used to recognize certain control state changes in a database gathered in two previous studies [R. J. Hanson, Proc. 9th Int. Congr. Acoust.1, 519 (1977), and R. J. Hanson, 7th Int. Congr. Phon. Sci. (1975)]. The results produced by the model compare very favorably with those produced by hand segmentation, indicating the viability of defining control states with distributional parameters.

ISSN:0001-4966    

DOI:10.1121/1.2004208

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

This paper is concerned chiefly with ray and normal mode models. The deterioration in the performance of ray models at low frequencies is illustrated with examples of shadow zones, sound channels, and caustics. As the frequency decreases, the acoustic properties of the bottom become increasingly important, and in constructing wave models for very low‐frequency applications it is necessary to take the properties of the bottom realistically into account. A normal mode model is used to show how the maximum depth to which the bottom properties must be known increases with decreasing frequency, and to illustrate the effects of shear waves. The paper concludes with a discussion of the validity of neglecting the coupling of compression and shear waves in homogeneous sediments.

ISSN:0001-4966    

DOI:10.1121/1.2004209

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

Following detonation of an underwater explosion, sound reverberates from the ocean surface and bottom and from scatterers within the medium itself. In addition, the ocean basin and obstructions within it, such as seamounts and islands, reradiate the acoustic energy incident upon them. This paper describes a computer model developed to predict the sound level that will ensue following detonation of a large underwater explosion as a result of this latter basin boundary reverberation. Experimental data from 10‐ton shots fired a number of years ago in the North Atlantic were used in the development of the computer model. The model has now been tested by comparing predictions with results obtained from CHASE XXI, a 660‐ton North Atlantic shot. This simplistic model has proved adequate for matching available experimental measurements in the North Atlantic Basin. Since the parameters on which the model is based are simply defined and applicable to any geometric configuration, the computations are expected to apply to other ocean basins as well. [Work supported by DNA.]

ISSN:0001-4966    

DOI:10.1121/1.2004214

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

A simple model has been derived for down‐slope enhancement (DSE) of underwater sound (the so‐called "Megaphone effect"). The model predicts DSE for shallow source/deep receiver travel paths across a sloping bottom as the difference between semispherical spreading (15 logR) + αRand cylindrical spreading (10 logR) + αRplus a constantK. In this formulation the frequency dependent attenuation term (αR) is canceled out by the subtraction operation leaving the simple formula: DSE = (5 logR) −K. By comparing measured values of DSE with those computed using this formula, the value of the constantKwas found to be about 5 dB when the rangeRis expressed in nautical miles. DSE for sound crossing the sloping sides of seamounts was found to be about 5 dB higher than that for continental slopes. These models have been used to compute the effective low‐frequency source levels for shipping lanes crossing the continental slopes in areas where the horizontal directionality of ambient noise has been measured. [Work supported by LRAPP.]

ISSN:0001-4966    

DOI:10.1121/1.2004217

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

Level recorders are widely used to measure the wide range analog level of noise and vibration. The conventional level recorder is sometimes inconvenient for the field measuring because of its size. With respect to the dynamic characteristics, this type recorder basically differs from the sound level meter conforming to ANSI and IEC standards. Especially in case of short duration signal input, there is large difference between them. The main reason is that the pen of the level recorder responds to the input at a constant speed, while the needle of a sound level meter deflects exponenially. The two newly developed level recorders have the pen response characteristics coinciding with the indication of a sound level meter, and is very convenient for use in the field accompanying a sound level meter because of its internal battery operation, smaller size, and lighter weight. One of these is specialized for the level recording adopting the dc servo mechanism, the other is for level and waveform recording adopting the high‐speed response multi‐pin‐head recording system.

ISSN:0001-4966    

DOI:10.1121/1.2004224

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

Sources of broadband noise in FC (forward curved blades) centrifugal fans were studied experimentally using a 0.28‐m‐diameter fan. Strong tangential and axial gradients of mean velocity and static pressure were found in the housing. Very large variations in velocity and velocity fluctuations occurred across the blade passages, from the suction surface to the adjacent pressure surface. Highly separated flow in the blade passages produced a jet‐wake flow pattern with large velocity fluctuations leaving the blades. Measurements on a narrow rotor (0.09 width‐to‐diameter ratio) mounted in a radial diffuser allowed the rotor only noise to be separated from the noise of the complete fan. Both the inlet and discharge rotor noise showed a continuous reduction in level with flow from maximum flow to the stall region. A miniature microphone mounted at various points on the blade surfaces showed a large increase in pressure fluctuations from the leading to the trailing edge of the blades. The fluctuations on the suction surface were much stronger than those on the pressure surface. Several high‐frequency spikes were noted in the pressure surface spectrum. Recommendations for new blade shapes and housing configurations are proposed to reduce the unsteady flows and the related broadband noise.

ISSN:0001-4966    

DOI:10.1121/1.2004229

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

A series of tests have been conducted to experimentally show the overall portable air compressor (PAC) noise escalation resulting from the degradation of design features intended to control air and structure‐borne PAC noise. These experiments were designed to determine the potential noise escalation associated with the degradation of each noise abatement feature individually and in combination with the other noise abatement features. Using information generated in the course of the investigation, a time‐dependent correlation has been developed showing the expected overall machine noise escalations associated with typical usage. The potential maximum and minimum noise escalations attained from varying degrees of machine maintenance were similarly determined. Test results have shown that an increase in noise as much as 18.5 dBAcan occur when the prime noise suppression design features of the PAC are degraded. A tape recording will be played to give a subjective appreciation of the level and character of noise escalation from the degradation of various sound control features. [Work supported by EPA.]

ISSN:0001-4966    

DOI:10.1121/1.2004231

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

The purpose of the present investigation was to examine the identification functions of a /so/‐/to/ continuum by normally hearing children and children with conductive and sensorineural hearing losses. Two six‐step continua were generated on the OVE III speech synthesizer at Haskins Laboratories. The continua differed only in the duration of friction. The friction duration of the first continuum ranged from 175 to 25 ms in 25‐ms steps while the friction duration of the second continuum ranged from 500 to 50 ms in 50‐ms steps. Normally hearing children listened to the short duration tape while the hearing impaired children listened to both continua. Results will be discussed relative to speech perception and the implication for articulation development.

ISSN:0001-4966    

DOI:10.1121/1.2004242

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP

摘要:

Twelve adult aphasics and 12 nonaphasic controls were tested for ability to discriminate (AX method) place differences between synthetic stop‐consonants as a function of formant transition duration. Two 11‐item arrays were generated, one with 40‐ms transitions and the second with 25‐ms transitions, and each array ranging in equal acoustic steps from [ba] through [da]to [ga]. Syllables were paired within these arrays at a two‐step difference level. The Iwo sets of stimulus pairs were intermixed for discrimination testing. Discrimination functions were computed by pooling response data across subjects within each subject group and stimulus set. Rank correlation coefficients (tau) were computed to compare discrimination functions within subject groups and to compare stimulus matched discrimination functions between subject groups. The within‐group comparisons revealed a greater association between the two functions within the control group than within the aphasic group. The between‐group comparisons revealed a greater association between the two functions obtained with 40‐ms transition stimuli. Formant transition duration significantly affected phonetic category boundries only in the aphasic group.

ISSN:0001-4966    

DOI:10.1121/1.2004247

出版商:Acoustical Society of America    

年代:1978

数据来源: AIP