REVIEWS OF ACOUSTICAL PATENTS Free

• GEORGE L. AUGSPURGER, Perception, Incorporated, Box 39536, Los Angeles, California 90039

• JOHN A. HOSSACK, Department of Biomedical, Electrical and Computer Engineering, University of Virginia, Charlottesville, Virginia 22908

• BENJAMIN MARKHAM, Acentech Incorporated, 33 Moulton Street, Cambridge, Massachusetts 02138

• LLOYD RICE, 11222 Flatiron Drive, Lafayette, Colorado 80026

• ERIC E. UNGAR, Acentech, Incorporated, 33 Moulton Street, Cambridge, Massachusetts 02138

Janet M. Rice et al., assignors to Interdev Technologies Inc.

26 June 2018; filed 10 February 2017

In spite of its title, this patent has potential applications other than speech recognition. It is essentially a processing method for improving the signal-to-noise ratio of a directional microphone array. In a given situation, the geometry of the array is known and the relative direction of the target source (a person talking) can be accurately predicted. If the array is steered toward the target, either physically or electronically, then sound from any other direction will be picked up by individual microphones with various degrees of phase shift that differ from on-axis phase shift, and can therefore be flagged and suppressed in subsequent processing.—GLA

Hong Gee Keong et al., assignors to TYMPHANY HK LIMITED

5 June 2018; filed 6 December 2016

Since a loudspeaker is really just an air pump, it should be possible to use the resulting air movement to help cool the speaker. The idea is attractive, and several patented designs exist. The design described here however is intended to cool associated electronic equipment rather than the speaker itself. Passive radiator 26 is driven by the fluctuating air pressure in back chamber 14. The movement of the passive radiator forces air through a second elongated chamber 28 whose sides serve as heat sinks for electronic circuit boards. This all sounds reasonable, but a passive radiator only operates over a narrow range of frequencies, perhaps one octave. It may work as intended if the speaker is a subwoofer.—GLA

Qing Miao et al., assignors to GOERTEK INC.

19 June 2018; filed 1 April 2015

A small thin audio device must utilize a small shallow loudspeaker. It is common practice to conduct sound from the front of the speaker through a thin duct that exits at the side. According to this patent, the configuration creates “redundant space” that can be put to use as an enlargement of the loudspeaker's rear chamber. Two practical embodiments are described.—GLA

YoungKyu Kim et al., assignors to HYUNDAI MOTOR COMPANY

19 June 2018; filed 2 November 2016

The goal of this patent is, “… to provide a speaker device for a vehicle capable of outputting an improved sound while allowing an enclosure to be compact.” The device takes the form of a shallow box that might be concealed below the floor or in a door panel. Sound from the front of the speaker is conducted through a vent to the interior of the vehicle. Sound from the rear may be vented into an adjacent hollow space. Although the patent shows construction details for the enclosure, it does not explain how the speaker can be accessed for service after it is installed in the vehicle.—GLA

Won Youl Bae et al., assignors to SAMSUNG ELECTRONICS CO. LTD.

5 June 2018; filed 7 August 2015

The underside of this elongated planar diaphragm speaker includes stiffening members to counteract bending and warping.—GLA

Wenxing Tu and Xiaoyan He, assignors to BEKEN CORPORATION

5 June 2018; filed 19 December 2016

This patent deals with a situation in which two “smart” loudspeakers (perhaps two cellular phones) receive and store audio files that must then be played in synchronization. The two files may be left and right channels of a stereo program for example. Since two clocks are involved, playback may not start at exactly the same time, and the files may not be played at exactly the same rate. The patent describes a method that is intended to predict and counteract the resulting time offsets. It involves analyzing a preliminary timing signal and then squeezing or stretching one audio file as required.—GLA

Sylvain J. Choisel and Adam E. Kriegel, assignors to APPLE INC.

5 June 2018; filed 1 June 2017

A “smart” home music player may be moved to different locations fairly often. For example, it may sit on a coffee table in front of a couch and then later be moved to a wall shelf, in which case the difference in nearby surfaces can make a pronounced difference in sound quality. It is common practice to counteract such room coloration with electronic equalization, and several methods for automatic room equalization have been patented. One interesting variant makes use of ordinary program material rather than a special test signal. It seems reasonable that this kind of equalizer could be augmented to sense physical or acoustical changes in the environment and then adjust equalization as required, with no input from the user. This Apple patent sets forth several practical embodiments of such a system.—GLA

Hidekatsu Yamada et al., assignors to KYOCERA CORPORATION

12 June 2018; filed 30 April 2013

For portable audio devices, using a vibrator to reproduce sound via bone conduction would seem to offer several advantages over conventional earphones, but it is difficult to achieve acceptable sound quality in practice. A few patented devices include both kinds of transducers so that bone conduction is only employed when the background noise level is high. A sensing microphone can be included so that switching between the two is done automatically. In addition, the position of the device can be sensed and used to double-check the choice of an appropriate transducer. This patent includes both methods plus an additional requirement for “space information,” i.e., global positioning system coordinates.—GLA

Dirk Jeroen Breebaart et al., assignors to Dolby International AB, Dolby Laboratories Licensing Corporation

19 June 2018; filed 18 April 2017

Object-oriented computer programming emerged more than 20 years ago but it has now become a hot topic in surround sound encoding. If a sound source is treated as an object, its associated metadata is encapsulated in a uniform, standardized code package. The metadata usually includes position, relative loudness, size, trajectory, etc. But suppose the source in question is diffuse and covers a very large area. This Dolby patent sets forth a method for achieving the desired goal when the “size” parameter exceeds a specified threshold value.—GLA

Glen A. Norris et al., assignors to Philip Scott, Lyren

12 June 2018; filed 17 November 2016

Imagine the following scenario when you wear headphones to receive a call on your smartphone: (a) Using an undisclosed process, your phone is somehow made aware of a particular physical object within your field of vision; (b) binaural processing is then applied to the incoming voice signal such that it seems to emanate from the chosen physical object. In less than 50 words, that summarizes this long patent. The procedure is rephrased at least three times in the patent Claims, and numerous bells and whistles are added in the descriptive text, but it is still hard to find anything significantly different from several other prior patents.—GLA

Richard J. Cartwright et al., assignors to Dolby Laboratories Licensing Corporation

19 June 2018; filed 5 April 2017

According to this Dolby patent, “The system may be applied in soundfield teleconferencing where a plurality of soundfields captured by teleconferencing devices equipped with soundfield microphones are to be presented to a remote listener wearing headphones.” Binaural signal processing is used to create virtual sound sources at arbitrary locations chosen to provide easy discrimination in a natural sounding environment.—GLA

Barry Lyn Holtzman

12 June 2018; filed 27 October 2017

Oscillations of a ski make it difficult for a skier to control the ski, particularly in aggressive skiing. A damper, attached to the ski in order to suppress these oscillations, consists in essence of an enclosure that is partially filled with a coarse aggregate of lead pellets and tungsten powder. The idea is that the aggregate will vibrate and impact against the enclosure walls and thus remove energy from the oscillations. Such dampers are added preferably where the oscillations are most severe, generally at the tip or at the back of a ski.—EEU

Esteban Martino Gonzalez et al., assignors to Airbus Operations S.L.

12 June 2018; filed 25 August 2016

This patent describes panel structures intended not for noise or vibration reduction, but for maintaining their basic structural integrity if they are subjected to impacts. A typical panel consists of a top sheet 17 and a bottom sheet 19 with honey-comb cell arrangements between them. The honey-comb arrangements consist of alternating portions, with one type 25 configured to expand laterally (that is in the direction perpendicular to the panel thickness) and a second type 35 configured to contract laterally in the presence of a localized impact D on the panel, thus enabling the structure to deform without major damage.—EEU

Mark A. Stebbins et al., assignors to GM Global Technology Operations LLC

26 June 2018; filed 27 July 2016

A toroidal damping assembly 38 is positioned between the upper and lower tubes of a telescoping steering wheel column. The damping assembly is basically a constricted flow device but it employs multiple orifices in one or more barriers between fluid reservoirs. The patent asserts that the design effectively damps numerous vibrational modes involving both vertical and lateral bending deflection.—GLA

Toby Scott Baudoin et al., assignors to KLX ENERGY SERVICES LLC

5 June 2018; filed 6 July 2015

The invention described in this patent is designed to be used in downhole equipment for oil and gas wells. We are told that a variety of vibration-generating tools may be attached to a pipe or coiled tubing string and used to perform various functions within the wellbore. An effective damping device must damp longitudinal vibrations without affecting rotational torque. The proposed design has a tubular housing filled with fluid, a movable piston that permits limited fluid flow around it, a spring that biases the piston toward the center of the cavity, and, “…mating profiles which rotationally lock said bottom sub and said tubular housing together but permit longitudinal movement therebetween.”—GLA

Emanuele Merlo et al., assignors to ALENIA AERMACCHI S.P.A.

12 June 2018; filed 3 December 2012

Aircraft engine noise is a perennial challenge. This patent is for a sound-absorbing panel element intended for engine housings (nacelles). It comprises a transondent sheet and a sound-reflective back-sheet, between which lies a honeycomb mesh of hollow cells and passages arranged so that sound resonates inside the cells.—BEM

James P. Stahl, Jr. et al., assignors to Specified Technologies, Inc.

12 June 2018; filed 15 May 2015

The top of a wall, where it meets the underside of the floor above, is a common source of acoustic leaks. This patent describes an approach to sealing the head of the wall, particularly for cases where resilient channels or other similar elements are used which separate one or more wall panels from the studs. A set of gaskets is applied to seal the gaps and maintain the partition's acoustical performance and fire rating.—BEM

Stephen Fung et al., assignors to Cisco Technology, Inc.

19 June 2018; filed 26 May 2016

The patent describes a method for scene identification to be performed by a hearing prosthesis device. Better information about the current audio environment would allow the device to perform compensation for, or adjustment to, certain more-or-less predictable aspects of the environment, thus providing a more accurate and useful output of the intended content.—DLR

Jeffrey E. Pierfelice et al., assignors to Toyota Motor Engineering & Manufacturing North America, Inc.

26 June 2018; filed 22 August 2015

This speech recognizer would use a partitioned model and would then gather specific information from the input speech to determine which parts of the speech model would be most useful for determining various vehicle control parameters, including engine controls, interior settings such as temperature, seat controls, radio settings, or ignition.—DLR

Vincent Pollet et al., assignors to Nuance Communications, Inc.

5 June 2018; filed 28 December 2016

The patented speech synthesis system would include the capability of producing speech output using a variety of speaking styles for various portions of the material. The system would include synthesis controls for various ages, genders, and a variety of situations, such as various emotions, news broadcast styles, dialog among friends, reading from written materials, among others.—DLR

Jean-Francois Vanreusel et al., assignors to Adobe Systems Incorporated

5 June 2018; filed 26 April 2016

The patented speech synthesis system would provide methods for pronouncing foreign words and phrases, such as place names, in any of a variety of speaking styles, to include regional dialects or specific accents. The system would include methods to allow the user to select a speaking style according to local customs or to personal preferences. A stored phonetic inventory would provide multiple pronunciations for each item.—DLR

Alistair D. Conkie et al., assignors to AT&T Intellectual Property I, L.P.

12 June 2018; filed 12 May 2014

The goal of this speech synthesis development system is to improve the prosodic pattern synthesis, particularly the fundamental frequency curve. Apparently, the patented method would consist of generating a large database of synthesized speech output and evaluating that output by comparison with natural speech, particularly geared toward locating comparable stress patterns.—DLR

Timothy Thomas Gray et al., assignors to Amazon Technologies, Inc.

5 June 2018; filed 28 June 2016

The patent describes a system for voice-activated retrieval of musical items, to be used, for example, in an automobile, in order to provide minimal driver distraction. Conversational strategies are discussed, such as help with the selection of particular items or efficiencies possible when particular items are requested frequently. The system would be able to use either local or network storage of material for playback.—DLR

William Chan et al., assignors to Google LLC

5 June 2018; filed 19 October 2017

This speech recognition system would use a sequence of neural networks to process an input speech signal, producing a transcription of the spoken input sequence. A first neural network would typically produce a sequence of tokens, such as acoustic elements. A second neural network would consist of an attention-based recurrent network, which would produce a series of scored substrings, typically using left-to-right beam search decoding.—DLR

Ran Achituv et al., assignors to VERINT SYSTEMS LTD.

5 June 2018; filed 24 October 2016

The patented speech recognition development method would consist of running a candidate speech recognizer on a particular large collection of speech materials to obtain transcriptions intended to represent the spoken content. Comparing the results would provide the information needed to build a language model. In other words, this seems to be a normal, standard approach to recognizer system development.—DLR

Sankalp Sandeep Modi

12 June 2018; filed 4 April 2017

The patent describes improvements to product and personal identifier systems, such as barcodes, RFID, magnetic stripes, smart cards, and fingerprint or iris readers. The improved system would make use of speech recognition techniques to allow a wider range of products to be managed using such systems. Various means are suggested that might help overcome some of the limitations of current speech recognition technology.—DLR

Timothy J. Grost et al., assignors to GM Global Technology Operations LLC

12 June 2018; filed 9 September 2015

The patented speech recognizer is particularly geared toward correct recognition of spoken names, including names of people and places, and is intended for use in a vehicle environment. The system would operate in connection with fixed services, such as speech-to-text, to provide a written version of a spoken name. Apparently, a text version of the name would be provided in the vehicle, allowing user confirmation of the recognition result.—DLR

Yifan Gong et al., assignors to Microsoft Technology Licensing, LLC

12 June 2018; filed 11 September 2015

The patent describes a speech recognition confidence classifier (CC), which is basically a standardized method for evaluating a recognition system performance. The CC achieves a result by processing a large database consisting of speech with transcriptions. The failure rates are then compiled into a standard form for system evaluation.—DLR

Shoji Hayakawa et al., assignors to FUJITSU LIMITED

26 June 2018; filed 24 October 2016

The patented speech recognizer would use a fairly typical front-end speech analyzer with a hidden Markov model to convert speech waveform frames into phoneme units. The phoneme units are then passed to a deep neural network classifier to determine the most likely word pattern represented by the observed phoneme sequence.—DLR

Haruki Kobayashi et al., assignors to CLARION CO., LTD.

26 June 2018; filed 25 February 2015

The vehicle speech recognition system described in the patent would perform an analysis of a user's speech input and convert the recognized result into an image to be displayed for the user. Vehicle control settings, including motor controls and parking brake, would be used to determine whether the user interaction would use speech or a touch pad and display unit.—DLR

Paul Mullen et al., assignors to General Electric Company

5 June 2018; filed 31 December 2013

A removable transducer housing is described for reducing variability in diagnostic ultrasound procedures performed on a patient's extremity (e.g., finger or toe). The housing, typically comprising a flexible material, is configured to encompass the patient's extremity, and possesses an aperture designed to securely hold the transducer in place during the procedure. The housing also has an aperture facing to the extremity through which imaging is performed.—JH

Chi Min Seow et al., assignors to Covidien LP

5 June 2018; filed 8 January 2014

In this invention, a diagnostic ultrasound transducer is paired with a laser light source so as to project onto the skin surface an illuminated line for guiding a surgical procedure. For example, a tumor may be visualized using ultrasound imaging and a laser illuminated line is used to identify on the tissue surface a “margin” around the tumor to guide safe and effective surgical resection.—JH

Nam-Woong Kim and Gil-Ju Jin, assignors to SAMSUNG MEDISON CO., LTD.

5 June 2018; filed 22 September 2014

This invention relates to an approach to automatically switching which of several transducers connected to a diagnostic ultrasound scanner is the intended active transducer. Rather than relying on the user to physically switch or connect a transducer, inactive connected transducers are given an intermittent pulse-echo test signal. If the addition of ultrasound gel is detectable, that is presumed to relate to the choice of the user to intend to use that transducer and the scanner switches the active transducer to correspond to the one on which a gel signal was just detected.—JH

Sei Kato and Hiroshi Hashimoto, assignors to General Electric Company

19 June 2018; filed 29 May 2015

A diagnostic ultrasound scanner creates a set of beamformed images using a range of assumed tissue speed of sound. Speed of sound is known to vary by tissue type but is not typically known a priori. The images reconstructed using the differing speeds of sound are analyzed using spatial frequency analysis. The image determined to be optimal is used to determine what is the optimal speed of sound to use and this speed is used in subsequent beamforming operations.—JH

Michael H. Slayton and Paul Jaeger, assignors to GUIDED THERAPY SYSTEMS LLC

26 June 2018; filed 17 November 2016

In this invention, ultrasound energy emitted from a diagnostic ultrasound transducer is monitored. In some instances, transducer temperature is also monitored. If the energy or temperature are determined to be out of a desirable range, the scanner controller changes the transmitted power or frequency in an attempt to bring the transmitted power to within an optimal working zone.—JH

Lucie Brasset et al, assignors to INSTITUT NATIONAL DE LA SANTE ET DE LA RECHERCHE MEDICALE-I.N.S.E.R.M., EDAP TMS FRANCE

26 June 2018; filed 23 October 2014

This invention relates to thermal treatment of tissues using application of focused ultrasound. The essence of the invention involves a computer-controlled transducer motion stage that sequences the probe location through an intended tissue target volume. The increments in motion are selected so that adjacent thermal treatment zones are juxtaposed. The invention also includes a compensation for attenuation due to target tissue depth.—JH