Nothing Special   »   [go: up one dir, main page]

WO2018021920A1 - Système de mesure de flux d'air en situation de parole sans masque - Google Patents

Système de mesure de flux d'air en situation de parole sans masque Download PDF

Info

Publication number
WO2018021920A1
WO2018021920A1 PCT/NZ2017/050104 NZ2017050104W WO2018021920A1 WO 2018021920 A1 WO2018021920 A1 WO 2018021920A1 NZ 2017050104 W NZ2017050104 W NZ 2017050104W WO 2018021920 A1 WO2018021920 A1 WO 2018021920A1
Authority
WO
WIPO (PCT)
Prior art keywords
sensor
airflow
user
input signals
nasal
Prior art date
Application number
PCT/NZ2017/050104
Other languages
English (en)
Inventor
Tom Gerard DE RYBEL
Donald James DERRICK
Original Assignee
The University Of Canterbury
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by The University Of Canterbury filed Critical The University Of Canterbury
Publication of WO2018021920A1 publication Critical patent/WO2018021920A1/fr

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
    • A61B5/087—Measuring breath flow
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
    • A61B5/087—Measuring breath flow
    • A61B5/0873—Measuring breath flow using optical means

Definitions

  • This invention relates to a speech airflow measurement system.
  • Speech airflow measurement devices are commonly used by speech therapists to diagnose and treat speech disorders.
  • the only way to measure airflow from speech precisely is to capture the air flow from the oral and nasal passages and pass it through a flow meter. To achieve accurate results, it is necessary to capture all the air flow and convey it through the flow meter.
  • Typical speech airflow recording techniques are based on pneumotachograph technology, which involves covering the mouth and/or nose with a mask and measuring air flow and air pressure differences across defined ports in the mask, or through connected tubes equipped with appropriate sensors. Other techniques involve either masks or sensory arrays that obscure the face of the speaker in order to collect speech airflow information.
  • breathing often takes significantly more effort, due to the constriction of the air flow in the mask ports and/or sensors. This has the potential to affect speech production and breathing patterns, as well as induce a sense of
  • Rothenberg also discloses a method for correcting recorded values of vowel nasalance by reducing the effect of crossover between the oral and nasal channels due to imperfect acoustic separation.
  • Rothenberg's method is intended for use with conventional speech airflow measurement devices which require an acoustic separation means held against the upper lip of a user. Such devices suffer from the disadvantages described above.
  • An additional or alternative object of at least preferred embodiments of the present invention is to at least provide the public with a useful alternative.
  • a speech airflow measurement system comprises a feature extraction module configured to receive input signals associated to a user from at least a first sensor and a second sensor, and determine an estimated shape and/or rate of airflow from at least part of the input signals.
  • At least some of the input signals from the first sensor are associated to a first airflow of the user.
  • the first airflow of the user comprises a nasal airflow.
  • at least some of the input signals from the second sensor are associated to a second airflow of the user.
  • the second airflow of the user comprises an oral airflow.
  • the input signals from the first sensor and/or second sensor include input signals associated to the first airflow of the user and input signals associated to the second airflow of the user, the feature extraction module configured to separate the input signals associated to the first airflow of the user from the input signals associated to the second airflow.
  • the input signals from the first sensor and/or the second sensor are associated to at least one frequency in the range 0.1 Hz to 70 Hz. In an embodiment the input signals from the first sensor and/or the second sensor are associated to at least one frequency in the range 20 Hz to 20 kHz.
  • the input signals from the first sensor and/or the second sensor are associated to at least one time period in the range 14ms to 10s.
  • the feature extraction module further comprises an oral airflow estimator configured to estimate oral airflow and/or a nasal airflow estimator configured to estimate nasal airflow.
  • the feature extraction module further comprises an oral audio estimator configured to estimate oral audio and/or a nasal audio estimator configured to estimate nasal audio.
  • the feature extraction module further comprises a nasalance module configured to compute at least one nasalance indicator from the input signals from the first sensor and/or from the input signals from the second sensor.
  • the feature extraction module further comprises a fundamental frequency (fO) tracker configured to extract fundamental frequency from the input signals from the first sensor and/or the input signals from the second sensor.
  • fO fundamental frequency
  • the feature extraction module further comprises a breath group tracker configured to compute at least one breath group from the input signals from the first sensor and/or from the input signals from the second sensor.
  • the feature extraction module further comprises a formant tracker configured to extract at least one vowel formant from the input signals from the first sensor and/or from the input signals from the second sensor.
  • the feature extraction module further comprises a voice quality tracker configured to extract at least one voice quality indicator from the input signals from the first sensor and/or from the input signals from the second sensor.
  • the feature extraction module is configured to receive input signals associated to the user from a third sensor not associated to either the first airflow of the user or the second airflow of the user.
  • system further comprises a pre-processing module configured to apply at least one correction to the input signals from the first sensor and/or to the input signals from the second sensor; and generate an output for use by the feature extraction module.
  • system further comprises a classification module configured to receive as input at least part of the feature extraction module output; and generate diagnostic information as an output.
  • system further comprises a post-processing module configured to receive as input at least part of the feature extraction output and/or at least part of the classification module output; and generate a visual representation as an output.
  • a post-processing module configured to receive as input at least part of the feature extraction output and/or at least part of the classification module output; and generate a visual representation as an output.
  • the speech airflow estimation system further comprises a headset comprising the first sensor positioned within at least a first airflow of the user; the second sensor positioned within at least a second airflow of the user; and a shielding member adapted to shield the first sensor from the second airflow, the shielding member adapted to provide an air gap between the shielding member and a face of the user while the headset is in use by the user.
  • the headset further comprises a sensor assembly comprising a nasal sensor assembly and an oral sensor assembly, the nasal sensor assembly including the first sensor, the oral sensor assembly including the second sensor.
  • the nasal sensor assembly and/or the oral sensor assembly include(s) respective Venturi structures shaped to case a constriction of a flow path within the nasal sensor assembly and/or the oral sensor assembly.
  • the headset further comprises the third sensor located substantially outside of the first airflow of the user and the second airflow of the user.
  • At least one of the sensors comprises a sound pressure sensor.
  • the at least one sensor may comprise a microphone, a strain gauge, or any other suitable sound pressure sensor.
  • at least one of the sensors comprises an airflow velocity sensor.
  • the at least one sensor may comprise a microphone, an anemometer, or any other suitable airflow velocity sensor.
  • At least one of the sensors comprises an airflow sensor.
  • At least one sensor may comprise a flow meter or an airflow velocity sensor in a constraining or partially constraining structure.
  • the input signal is filtered so that the estimation of rate of airflow is based on 'pseudo-sound' signals.
  • the input signal may be filtered using an inverse filter.
  • the rate of airflow is estimated using input signals from the at least one sensor combined with an audio signal.
  • a method of estimating speech airflow comprises receiving input signals associated to a user from at least a first sensor and a second sensor, and determining an estimated shape and/or rate of airflow from at least part of the input signals. It is intended that reference to a range of numbers disclosed herein (for example, 1 to 10) also incorporates reference to all rational numbers within that range (for example, 1, 1.1, 2, 3, 3.9, 4, 5, 6, 6.5, 7, 8, 9 and 10) and also any range of rational numbers within that range (for example, 2 to 8, 1.5 to 5.5 and 3.1 to 4.7) and, therefore, all sub-ranges of all ranges expressly disclosed herein are hereby expressly disclosed. These are only examples of what is specifically intended and all possible combinations of numerical values between the lowest value and the highest value enumerated are to be considered to be expressly stated in this application in a similar manner.
  • This invention may also be said broadly to consist in the parts, elements and features referred to or indicated in the specification of the application, individually or collectively, and any or all combinations of any two or more said parts, elements or features, and where specific integers are mentioned herein which have known equivalents in the art to which this invention relates, such known equivalents are deemed to be incorporated herein as if individually set forth.
  • a component may be, but is not limited to being, a process running on a processor, a processor, an object, an executable, a thread of execution, a program, and/or a computer.
  • an application running on a controller and the controller can be a component.
  • One or more components may reside within a process and/or thread of execution and a component may be localized on one computer and/or distributed between two or more computers.
  • connection to' as used in this specification in relation to data or signal transfer includes all direct or indirect types of communication, including wired and wireless, via a cellular network, via a data bus, or any other computer structure. It is envisaged that they may be intervening elements between the connected integers. Variants such as 'in communication with', 'joined to', and 'attached to' are to be interpreted in a similar manner. Related terms such as 'connecting' and 'in connection with' are to be interpreted in the same manner.
  • Figure 1 shows an example of a signal processing system adapted to receive and process speech airflow signals
  • Figure 2 shows examples of modules implemented in the computing device of figure 1 ;
  • Figure 3 shows an example of the feature extraction module from figure 2;
  • Figure 4 shows an example of a computational technique for separating oral and nasal signals;
  • Figure 5 shows an example of audio and airflow estimation
  • Figure 6 shows an example of nasalance
  • FIG 7 shows an example of fundamental frequency (fO).
  • Figure 8 shows an example of a headset adapted for use with the speech airflow estimation system of figures 1 to 3; and Figures 9 and 10 show the sensor assembly from figure 8 in more detail.
  • Described below is a system for estimating speech airflow based on pressure and/or velocity measurements that does not require a mask and does not require a physical airflow separator that contacts the face of the user.
  • Speech airflow comprises two types of flow - laminar and turbulent.
  • Laminar airflow moves in parallel layers with no interaction between the layers.
  • Turbulent airflow involves chaotic property changes, with low momentum diffusion and high momentum convection. This means that turbulent airflow moves in vortices that mix the air up, and so moves away from the source more slowly than laminar airflow.
  • Sound is a longitudinal wave that propagates through air during speech. Sound produced by speech is usually a mixture of periodic and aperiodic longitudinal motion that ranges in frequency from about 20 Hz to about 20,000 Hz. Most of the sound energy from speech occurs in the frequency range of about 80 Hz to 8,000 Hz, depending on the age, body size and physical gender of the speaker.
  • the airflow contribution to pressure and velocity changes detected by a sensor can be described as 'pseudo-sound'.
  • Pseudo-sound is produced by the interaction of outgoing airflow from the mouth and nose interacting with the sensor.
  • the laminar airflow will interact with the sensor in a relatively continuous fashion, and the turbulent airflow will cause random fluctuations in the sensor response.
  • true sound is a longitudinal wave that will create fluctuations in the sensor according to the frequency of the sound waves produced.
  • speech airflow consists of a relatively large bulk volume of air molecules moving in, or out, of the mouth and nose.
  • speech sound is a longitudinal compression wave, which yields no net mass motion.
  • the directionality and mass generated due to speech airflow means that speech airflow generates higher intensity activity at much lower frequencies than speech sound.
  • a common embodiment of an air pressure or velocity sensor is a microphone.
  • Microphones typically comprise filters that supress low frequency information below about 20 Hz (known as 'high-pass' filtering). Their associated electronic circuits often incorporate filters to further suppress low frequencies. Low frequency information introduces artefacts into the recording, such as rubbing sounds and microphone 'pops', and reduces the amplitude difference that can be successfully recorded. Such artefacts can be introduced by a windy environment, an environment where an external object might touch or shake the microphone, or when a speaker breathes close to the microphone. High-pass filtering is standard practice, as there is little useful speech audio below 20 Hz and human hearing has little sensitivity for such low frequencies.
  • the low frequency information that is usually filtered out of audio recordings has the potential to provide useful information about speech airflow.
  • speech airflow can be estimated.
  • Figure 1 shows an example of a signal processing system adapted to receive and process speech airflow signals.
  • the operating environment of figure 1 is an example of a suitable operating environment. It is not intended to suggest any limitation as to the scope of use or functionality of the operating environment.
  • Example computing devices include, but are not limited to, personal computers, server computers, hand-held or laptop devices, mobile devices, multiprocessor systems, consumer electronics, mini computers, mainframe computers, and distributed computing environments that include any of the above systems or devices.
  • mobile devices include mobile phones, tablets, and Personal Digital Assistants (PDAs).
  • embodiments are described in the general context of 'computer readable instructions' being executed by one or more computing devices.
  • computer readable instructions are distributed via tangible computer readable media.
  • computer readable instructions are implemented as program modules. Examples of program modules include functions, objects, Application
  • FIG. 1 Shown in figure 1 is a system 100 comprising a computing device 105 configured to implement one or more embodiments described below.
  • computing device 105 includes at least one processing unit 110 and memory 115.
  • memory 115 is volatile (such as RAM, for example), non-volatile (such as ROM, flash memory, etc., for example) or some combination of the two.
  • a server 120 is shown by a dashed line notionally grouping processing unit 110 and memory 115 together.
  • computing device 105 includes additional features and/or
  • additional storage including, but not limited to, magnetic storage and optical storage.
  • additional storage is illustrated in figure 1 as storage 125.
  • computer readable instructions to implement one or more embodiments provided herein are maintained in storage 125.
  • storage 125 stores other computer readable instructions to implement an operating system and/or an application program.
  • Computer readable instructions are loaded into memory 115 for execution by processing unit 110, for example.
  • Memory 115 and storage 125 are examples of computer storage media.
  • Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVDs) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by computing device 105. Any such computer storage media may be part of device 105.
  • computing device 105 includes at least one communication connection 140 that allows device 105 to communicate with other devices.
  • the at least one communication connection 140 includes one or more of a modem, a Network Interface Card (NIC), an integrated network interface, a radio frequency
  • transmitter/receiver an infrared port, a USB connection, or other interfaces for connecting computing device 105 to other computing devices.
  • the communication connection(s) 140 facilitate a wired connection, a wireless connection, or a combination of wired and wireless connections.
  • Communication connection(s) 140 transmit and/or receive communication media .
  • Communication media typically embodies computer readable instructions or other data in a "modulated data signal” such as a carrier wave or other transport mechanism and includes any information delivery media.
  • modulated data signal includes a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal.
  • device 105 includes at least one input device 145 such as keyboard, mouse, pen, voice input device, touch input device, infrared cameras, video input devices, and/or any other input device.
  • device 105 includes at least one output device 150 such as one or more displays, speakers, printers, and/or any other output device.
  • Input device(s) 145 and output device(s) 150 are connected to device 105 via a wired connection, wireless connection, or any combination thereof. In an embodiment, an input device or an output device from another computing device is/are used as input device(s) 145 or output device(s) 150 for computing device 105.
  • components of computing device 105 are connected by various interconnects, such as a bus.
  • interconnects include one or more of a Peripheral Component Interconnect (PCI), such as PCI Express, a Universal Serial Bus (USB), firewire (IEEE 13104), and an optical bus structure.
  • PCI Peripheral Component Interconnect
  • USB Universal Serial Bus
  • IEEE 13104 Firewire
  • optical bus structure an optical bus structure.
  • components of computing device 105 are interconnected by a network.
  • memory 115 in an embodiment comprises multiple physical memory units located in different physical locations interconnected by a network.
  • storage devices used to store computer readable instructions may be distributed across a network.
  • a computing device 155 accessible via a network 160 stores computer readable instructions to implement one or more embodiments provided herein.
  • Computing device 105 accesses computing device 155 in an embodiment and downloads a part or all of the computer readable instructions for execution.
  • computing device 105 downloads portions of the computer readable instructions, as needed .
  • some instructions are executed at computing device 105 and some at computing device 155.
  • a client application 165 enables a user experience and user interface.
  • the client application 165 is provided as a thin client application configured to run within a web browser.
  • the client application 165 is shown in figure 1 associated to computing device 155. It will be appreciated that application 165 in an embodiment is associated to computing device 105 or another computing device.
  • Shown in figure 1 is a first sensor 170 and a second sensor 175.
  • the sensors are connected to computing device 105 via a wired connection, wireless connection, or any combination thereof.
  • the first sensor 170 is positioned within at least a first airflow of a user.
  • a first airflow is nasal airflow.
  • the signals generated by the first sensor 170 are associated to a nasal airflow of a user.
  • the second sensor 175 is positioned within at least a second airflow of a user.
  • a second airflow is oral airflow.
  • the signals generated by the second sensor 175 are associated to an oral airflow of a user.
  • the first sensor 170 and the second sensor 175 each comprise respective microphones configured to detect airflow pressure of a sound in a pseudo- sound frequency range of about 0.1 to 70 Hz.
  • the term ow frequency' or 'pseudo-sound' denotes this frequency range of about 0.1 to 70 Hz.
  • the first sensor 170 and the second sensor 175 each comprise respective microphones configured to detect airflow pressure of a sound in a frequency range of about 20 Hz to about 20 kHz. This is in what is generally considered the audible range.
  • Acoustic microphones have the potential to be suitable sensors for capturing nasal and oral airflow information.
  • Microphones are mass-produced and compact.
  • the low frequency response of most commercial microphones is insufficient as they typically have built in high-pass filters.
  • This shortcoming can be improved with the use of an inverse filter, designed to compensate to a useful extent for the loss of microphone sensitivity and associated phase shift due to the acoustic high-pass filter in the microphone and/or the microphone pre-amplifier.
  • microphones without a high-pass filter, or a suitable high-pass filter is preferred to maintain the air flow information in the signal.
  • a suitable high-pass filter is one so arranged as to pass information above about 0.1 Hz, while rejecting lower frequency (very slow) pressure variations, such as those caused by atmospheric pressure variations, static pressure from building ventilation systems, changes in altitude, and so on, that would create large offsets in the microphone output unrelated to the airflow information.
  • the high-pass filter comprises a vented chamber in the device, one side of it the microphone membrane, where the venting port in this chamber to the outside sets the high-pass cut-off frequency.
  • the high-pass filter forms part of the pressure equalization system.
  • Suitable types of acoustic microphone include MEMs, electret microphones (self- polarized, capacitive), condenser microphones (externally-polarized, capacitive), other electret/ condenser microphones (for example Sennheiser's modulated bridge design), piezo-electric microphones, dynamic (moving-coil) inductive microphones, optical microphones, and interferometer microphones.
  • strain gauges of the resistive type in all common configurations, such as full bridge, single element, half-bridge, and all their modes of excitation
  • strain gauges of the capacitive type in all their common configurations, and all their modes of excitation
  • thermal sensors such as hot wire anemometers, ultrasonic (acoustic) flow meters, optical flow meters based on transmitted or reflected light from particles in the air flow (eg : water droplets), interferometers, and mechanical sensors, such as (rotating) vanes, oscillating structures, vortex shedding, and in-flow deflection sensors.
  • thermal sensors such as hot wire anemometers, ultrasonic (acoustic) flow meters, optical flow meters based on transmitted or reflected light from particles in the air flow (eg : water droplets), interferometers, and mechanical sensors, such as (rotating) vanes, oscillating structures, vortex shedding, and in-flow deflection sensors.
  • Other types are based on measuring pressure differences across a constriction in the flow path, such as a venturi, or an acoustic impedance in the flow path, such as a fine wire mesh.
  • the computing device 105 is adapted to receive signals from the first sensor 170 and/or second sensor 175.
  • the computing device is adapted to process the signals and generate output for an output device 150 for example a display.
  • Figure 2 shows examples of modules implemented in one embodiment of the computing device 105. It will be appreciated that in some embodiments not all modules shown in figure 2 are implemented on computing device 105. It will also be appreciated that in some embodiments additional modules are implemented on computing device 105. Shown in figure 2 is a pre-processing module 200 adapted to apply corrections to the signals from the first sensor 170 and/or second sensor 175. As described above, the first sensor 170 is associated to nasal airflow and the second sensor 175 is associated to oral airflow. In an embodiment the pre-processing module 200 is further configured to generate an output for use by the feature extraction module 205.
  • the sensor signal from the sensors 170 and 175 is pre-processed in the pre-processing module 200 and at least one transfer function of the sensors and their interface corrected to achieve a wider useful frequency response.
  • the pre-processing module 200 is adapted to correct for other short-comings of the sensor and/or signal path, including but not limited to: recovering from sensor overload conditions due to plosives (short but intense bursts of air during eg sounds), DC baseline drift, and time-aligning the signals to compensate for potentially unequal filter delay as well as energy propagation delays between the mouth/nose and oral/nasal sensors. These delays may be frequency-dependent.
  • the pre-processing module 200 is adapted to correct for undesirable cross-coupling between the oral and nasal signals in the sensor signals. To minimise this undesirable cross-coupling, good placement of the sensors, and suitable physical structures (such as the shielding structures) positively impact the degree of separation between oral and nasal information in the sensor signals.
  • the pre-processing module 200 uses data from the first sensor 170 to filter data from the data recorded by the second sensor 175.
  • the preprocessing module 200 in an embodiment uses data from the second sensor 175 to filter data from the data recorded by the first sensor 170. Any combination of using data from at least one sensor to filter data from at least one other sensor is possible in various embodiments.
  • the third sensor 180 is configured to reduce background noise interference.
  • a feature extraction module 205 is configured to receive input signals from sensor 170 and/or sensor 175.
  • a feature extraction module 205 is configured to receive pre-processed input signals from sensor 170 and/or sensor 175 via pre-processing module 200.
  • the feature extraction module is configured to receive input signals directly from sensor 170 and/or sensor 175.
  • the first sensor 170 is associated to nasal airflow and the second sensor 175 is associated to oral airflow.
  • a third sensor 180 is positioned outside the nasal airflow captured by the first sensor 170 and outside the oral airflow captured by the second sensor 175.
  • the third sensor 180 is configured to obtain an audio signal that is substantially free from any artefacts due to any pseudo sound impacting the first sensor 170 and/or second sensor 175.
  • the third sensor 180 has the potential to be used for noise reduction purposes in circumstances where the system is used in less than ideal conditions.
  • the feature extraction module 205 is further configured to calculate estimated rate of airflow from at least part of the signals from the sensor(s) 170 and 175 using one or more mathematical methods.
  • the airflow is estimated from the frequency range in which most of the non-sound airflow energy impacts the sensors.
  • the frequency from which airflow is estimated is in the range 0.1 Hz to 70 Hz. The estimation is made within a time period of 14 ms to 10 s.
  • An embodiment of the speech airflow estimation technique takes as input the force from the airflow applied to a pressure or velocity or flow transducer. This provides a reasonable approximation of airflow change over time that relates well in time and space to individual speech sounds, or phones, as well as words, phrases, or speech breath cycles.
  • a classification module 210 is adapted to take input from the feature extraction module 205 and is adapted to output diagnostic information.
  • Input obtained from the feature extraction module 205 includes nasalance (a measure of the nasality of speech), fundamental frequency (fO) tracking (a measure corresponding to perceived pitch), and vowel formant tracking (a measure of vowel quality).
  • the classification module 210 is adapted to identify nasal airflow in speech and/or oral airflow in speech. In an embodiment, the classification module 210 is adapted to capture physical speech impairment symptoms and/or chemical speech impairment symptoms. This includes but is not limited to, hyper or hypo-nasality, incorrect oral and nasal airflow patterns, incorrect prosody (as measured in part by fundamental frequency tracking, and in part by speech timing), incorrect tone (as measured by the fundamental frequency tracking), and incorrect pronounciation or word choice (as captured through the audio portion of the signal.)
  • the classification module 210 is used to identify one or more of speech breathing irregularities, phonation irregularities, speech production irregularities, velopharyngeal inadequacy from any cause, lip closure inadequacy from any cause, language and accent.
  • Shown in figure 2 is a post-processing module 215 adapted to store the processed data.
  • the post-processing module 215 is adapted to receive input data from the classification module 210 and/or from the feature extraction module 205.
  • the post- processing module 215 is adapted to receive input data directly from the feature extraction module 205 in embodiments that do not include a classification module 210.
  • the post-processing module 215 stores, in at least one data file, one or more of: raw sensor information, audio, nasal airflow, oral airflow, and/or computed quantities from this information such as nasalence, fundamental frequency or any combination of the foregoing.
  • the post-processing module 215 generates a visualisation or visual representation of the raw sensor information, audio, nasal airflow, and/or oral airflow output, and/or computed quantities from this information such as nasalence,
  • the visualisations are displayed for example on an output device 150 from figure 1 such as a display.
  • the post-processing module 215 is adapted to display one or more of identifying markers for physical speech impairment symptoms, identifying markers for chemical speech impairment symptoms, identifying markers for speech disability diagnosis, including but not limited to speech breathing issues, phonation issues, or acute speech production issues, identifying markers for velopharyngeal inadequacy from any cause, identifying markers for lip closure inadequacy from any cause, identifying markers for identifying language and accent.
  • the post-processing module 215 is adapted to provide a biofeedback tool for purposes including but not limited to, language training, accent training, or training for people with speech and/or hearing disabilities.
  • Figure 3 shows an example embodiment of the feature extraction module 205 from figure 2.
  • the feature extraction module 205 includes an oral airflow estimator 300 configured to estimate oral airflow and/or a nasal airflow estimator 305 configured to estimate nasal airflow.
  • the oral flow estimator 300 and/or nasal flow estimator 305 includes at least one filter configured to estimate airflow.
  • filters include low-pass filters and band-pass filters.
  • the oral flow estimator 300 and/or nasal flow estimator 305 are configured in various embodiments to estimate airflow based at least partly on one or more of: zero-crossing rates, periodicity, an autocorrelation, an instantaneous frequency, a frequency energy, a statistical measure, a rate of change, an intensity root mean square value, time spectral information, a filter, a filter bank, a demodulation scheme, time window gating, formant tracking, or the acoustic signal itself.
  • the oral flow estimator 300 and/or nasal flow estimator 305 are configured to estimate airflow based at least partly on one or more of: heuristics, logic systems, mathematical analysis, statistical analysis, learning systems, gating operation, range limitation, and normalization on the candidate air flow portion.
  • the rate of airflow is estimated using input signals from the sensor 170 and/or sensor 175 combined with an audio signal.
  • airflow extraction is accomplished by filtering out the audio signal, leaving the nasal and/or oral airflow.
  • An oral audio estimator 310 is configured to obtain oral audio from the oral signal.
  • a nasal audio estimator 315 is configured to obtain nasal audio from the nasal signal.
  • the oral and/or nasal audio is filtered so that the audio portion is based on a reasonable acoustic range perceived by humans.
  • the oral audio estimator 310 and/or nasal audio estimator 315 includes at least one filter configured to estimate airflow. Examples of filters include high-pass filters and band-pass filters.
  • one or more of the estimation filters is configured so that their group delays (frequency-dependant time delay) are corrected for and the time-domain signals obtained after filtering for airflow and/or audio are time-aligned.
  • the estimation filters are complemented with group delay (frequency- dependent time delay) compensation filters so the time-domain signals obtained after filtering for airflow and/or audio are time-aligned.
  • a fundamental frequency (fO) tracker 320 is configured to extract fundamental frequency from the signal.
  • the fundamental frequency is related to pitch perception. FO variations contribute to prosody and in tonal languages help distinguish lexical categories.
  • the fO tracker has the potential to provide the system with uses including language accentedness training and learning of tonal languages.
  • a voice quality tracker 325 is configured to extract at least one voice quality indicator from the signal.
  • the voice quality tracker 325 is configured to apply one or more of harmonic spacing, harmonic ratios, jitter, shimmer, harmonicity.
  • a formant tracker 330 is configured to extract at least one vowel formant from the signal.
  • the formant tracker 330 is configured to implement linear predictive coding.
  • a breath group tracker 335 is configured to extract at least one breath group from the signal.
  • the breath group tracker 335 is configured to implement sound/silence tracking.
  • a nasalance module 340 is configured to compute at least one nasalance indicator from the oral and nasal signal channels. In an embodiment, nasalance is computed at least partly from amplitude of the nasal signal divided by the amplitude of the oral signal plus the amplitude of the nasal signal. In an embodiment, nasalance is computed using a nasalance formula filtered to exclude extraneous data such as airflow.
  • the nasalance module has the potential to provide the system with clinical uses for cleft palate and deaf speech, as well as accentedness training.
  • Figure 4 shows an example of a computational method for separating oral and nasal signals.
  • the method comprises an adaptive filter-based technique.
  • the source separation method at least partially undoes the mixing of the oral and nasal signals at each sensor by estimating the mixing coefficients.
  • the mixing coefficients can be simplified to constants, or treated as frequency-dependent.
  • the method further comprises at least partially undoing the linear process of summation that occurs at each sensor.
  • the flow and acoustic separation is never perfect in practice, therefore some correction may be required or at least advantageous.
  • the coefficients for the methods are obtained by a calibration technique, where the user follows a specific procedure and generates oral and/or nasal cues accordingly.
  • varying degrees of blind source separation are employed.
  • the method using for instance cross-correlations between the signals, automatically tunes adaptive filters to remove the contribution of one flow to the other using optimization techniques.
  • One example of an optimisation technique is a gradient descent algorithm such as least-squares. It is advantageous to control the filter adaptation process using a voice activity detection scheme, so that the adaptation only proceeds during those portions of speech where useful signals are present.
  • a goal of the method 400 shown in figure 4 is to measure the real oral and/or nasal signals associated for example to sensors 170 and 175 respectively.
  • part of the signal from 405 is summed to the signal 410, yielding an actually measured signal 415.
  • the signal 405 is modified by its passage through the air and mechanical structures, a process which is represented by an unknown filter 420 and an unknown delay 425.
  • the signal part of the signal from 410 is summed to the signal 405, yielding the actually measured signal 430.
  • the signal 410 is modified by its passage through the air and mechanical structures, a process which is represented by an unknown filter 435 and an unknown delay 440.
  • inverse filters are implemented that undo the summation process.
  • estimates of the real oral and nasal signals are obtained through subtracting inverse-filtered versions of the mixed signals.
  • adaptive filters 455 and 460 are used.
  • signal 430, minus the estimated contribution of 410 becomes signal 445, which is an estimate of 405.
  • signal 415, minus the estimated contribution of 405 becomes signal 450, which is an estimate of 410.
  • this process exhibits an iterative convergence as adaptive filters 455 and 460 adapt to minimize the cross-coupling.
  • the inverse filters 455 and 460 are obtained for example by minimizing cross- correlations through a gradient descent technique, such as least-squares. As the oral and nasal signals have quite different wave shapes, their natural cross-correlation is below unity, and a degree of separation using such a method is possible.
  • Figures 5 to 7 show examples of biofeedback displayed for example on the output device 150 of figure 1.
  • Figure 5 shows an example of audio and airflow estimation from an ideal speech example shown at 500. Airflow for the estimation is shown at 505. An audio waveform is shown for example at 510.
  • An example of speech obtained from a user is shown at 520. Airflow for the speech sample is shown at 525. An audio waveform is shown for example at 530.
  • Figure 6 shows an example of nasalance. Nasalance from an ideal speech example is shown at 600. Nasalance recorded from a user is shown at 605.
  • Figure 7 shows an example of fundamental frequency (fO). Fundamental frequency from an ideal speech example is shown at 700. Fundamental frequency recorded from a user is shown at 705.
  • fO fundamental frequency
  • Figure 8 shows an example of a headset 800 adapted for use with the speech airflow estimation system described above.
  • Headset 800 comprises a boom-arm 805 rotatably connected to a headband 810 via hinge 815. Also shown adjacent the hinge 815 is a pad configured to enclose an ear of the user. It will be appreciated that this pad is an optional feature of the headset 800.
  • a sensor assembly 820 is mounted at an end of the boom-arm 805 distal from the hinge 815.
  • the sensor assembly 820 comprises a nasal sensor assembly 825 and an oral sensor assembly 830.
  • the nasal sensor assembly 825 is elevated with respect to the oral sensor assembly 830.
  • the sensor assembly 820 includes the first sensor 170 mounted within the nasal sensor assembly 825 at least partially within a nasal flow of the user.
  • the sensor assembly 820 further includes the second sensor 175 mounted within the oral sensor assembly 830 at least partially within an oral flow of the user.
  • a communication cable 835 connects the headset 800 to other components of the airflow estimation system, for example computing device 105.
  • the connection between the headset 800 and computing device 105 is wireless thereby obviating the need for a communication cable 835.
  • the communication cable 835 includes the sensor interface
  • the headset 800 further comprises the third sensor 180 adapted to record audio.
  • a third sensor is a microphone.
  • the third sensor 180 is located on the boom arm 805 substantially outside of the sources of oral and nasal airflow. In an embodiment the third sensor 805 is located substantially within the source of oral and/or nasal flow, for example within the sensor assembly 820.
  • the third sensor 180 is contained within a shielding device adapted to shield the third sensor 180 from low frequency signals. Alternatively or additionally the output from the third sensor 180 is filtered to suppress low frequency signals.
  • Figure 9 and Figure 10 show different views of an embodiment of the sensor assembly 820 in more detail.
  • the sensor assembly 820 comprises a nasal sensor assembly 825 and an oral sensor assembly 830.
  • Nasal sensor assembly 825 includes first sensor 170.
  • Oral sensor assembly 830 includes second sensor 175.
  • the sensor assembly 820 includes a shielding member 900.
  • the shielding member 900 is shaped to include a first planar surface and a second planar surface.
  • the nasal sensor assembly 825 is mounted to the first planar surface.
  • the oral sensor assembly 830 is mounted to the second planar surface.
  • the shielding member 900 is shaped and positioned so as to provide a partial separation between a nasal airflow of the user and an oral airflow of the user.
  • the shielding member 900 is shaped and positioned so as to shield the first sensor 170 within the nasal sensor assembly 825 and/or the second sensor 175 within the oral sensor assembly 830 from undesired airflows.
  • the shielding member 900 is further shaped and positioned so as to provide a mount to attach the sensor assembly 820 to the boom arm 805.
  • the nasal sensor assembly 825 includes a pair of side walls 905 and 910 defining a cavity between side walls 905 and 910.
  • the cavity extends through the nasal sensor assembly 825.
  • the width of the cavity is non-uniform along a length of the cavity.
  • the oral sensor assembly 830 includes a Venturi-like structure similar to that of the nasal sensor assembly 825.
  • the oral sensor assembly 830 includes a pair of side walls 930 and 935 defining a cavity between side walls 930 and 935.
  • the cavity extends through the oral sensor assembly 830.
  • the width of the cavity is non-uniform along a length of the cavity.
  • a pair of extensions 940 and 945 extend within the cavity to form a Venturi-like structure to which the second sensor 175 is connected through an aperture 950.
  • the Venturi structures of the nasal sensor assembly 825 and/or the oral sensor assembly 830 are configured to at least partially reduce the air pressure at the sensors 170 and 175 to better match the dynamic range of the microphone to the high pressure events due to (pseudo) speech airflow, funnel the air flow from the nose and mouth to the sensors 170 and 175, and/or generate a degree of turbulence that makes flow more easily identified and audible.
  • the Venturi structures are shaped to cause a constriction of the flow path within the nasal sensor assembly 825 and/or the oral sensor assembly 830, which causes the pressure to decrease, but the flow velocity to increase.
  • the Venturi structures reduce the pressure of the airflow locally, to better match the pressure of this flow to the dynamic range of the sensor.
  • the Venturi structures also serve to constrain and duct part of the flow, provide separation between the oral and nasal flows, and ensure a more turbulent nature of the flow, which is easier to capture using some types of sensors, such as microphones.
  • the Venturi structures have only one explicit sensing port, in the constricted part of the Venturi structure.
  • the second port is the environment, which is shared by the sensor reference port, for example the breathing port in an acoustic microphone to the environment.
  • Two Venturi structures are used to arrive at a convenient mechanical assembly incorporating the sensor elements.
  • At least one shielding device is adapted to perform frequency- selective shielding.
  • the frequency-selective shielding allows low frequency signals and attenuates high frequency signals.
  • the frequency-selective shielding allows high frequency signals and attenuates low frequency signals.
  • the aperture 925 is formed in the nasal sensor assembly 825 at a location within the assembly 825 corresponding to the narrowest width of the cavity of the assembly 825.
  • the aperture 950 is formed in the oral sensor assembly 830 at a location within the assembly 830 corresponding to the narrowest width of the cavity of the assembly 830.
  • the sensors 170 and 175 are positioned so as to measure airflow though the apertures 925 and 950 respectively.
  • the width of the nasal assembly side wall 905 and nasal assembly side wall 910 is greater at a lower portion of the nasal sensor assembly 820 than at an upper portion. This difference in relative lengths, combined with the shielding member 900, serves to capture nasal airflow more effectively. Nasal flow enters largely from above into the nasal sensor assembly 825.
  • the shielding member 900 at least partially channels the nasal flow into the nasal sensor assembly 825.
  • the nasal sensor assembly 825 and the oral sensor assembly 830 are positioned relative to each other so that the separation of oral and nasal flow paths is improved, and also the propagation path of the oral and nasal flows to the respective sensor structures is made more equal.
  • the oral sensor assembly 830 is set back from the face in relation to the nasal sensor assembly 825.
  • the lips of a user tend to protrude from the face meaning that a set back is desirable.
  • Another factor is that pursed lips should preferably not touch any part of the sensor assembly 820 for comfort, but also to reduce rubbing noises.
  • the nasal sensor assembly 825 is shaped with a width such that most of the airflow from the nostrils is captured.
  • the oral sensor assembly 830 is shaped with a height such that a sufficient portion of the airflow from the mouth, even when lip positions shift during speech production. In an embodiment, these relative dimensions are also somewhat over-sized to help accommodate differences in the relative placement of the sensor assembly 820 with respect to the user's face, as well as various facial geometries.
  • the sensor assembly 820 is positioned so that it does not touch the face of the user at rest and during speech. This has the potential to improve the comfort for the user, prevent unnatural speech patterns, and prevent structure-born interference noises in the sensor data due to rubbing and impacting of the sensor assembly with the face and lips during use.
  • nasal sensor assembly 825 and oral sensor assembly 830 form a duct which further shields their respective sensors 170 and 175 from spurious airflow, as well as provide some degree of uniformity in the airflow that is captured during speech, by offering a defined orifice that funnels a defined portion of the respective flows to the sensors 170 and 175.
  • sensor assembly 820 comprising a nasal sensor assembly 825 and oral sensor assembly 830 is one example of an assembly.
  • a sensor assembly includes a first sensor, a second sensor, and optionally a small separator/shielding structure. The first sensor and second sensor are positioned relative to the nose and lips of a user to achieve some degree of initial separation.
  • the system described above has the potential to demonstrate nasality to a user in real-time.
  • the user is able to use this biofeedback to change the amount of nasality in the user's speech.
  • nasality biofeedback is accent modification. By making the user aware of the oral and nasal production patterns, speakers can be trained to produce the desired patterns of oral and nasal airflow in speech.
  • nasality biofeedback is voice training (singing lessons), where the user can gain information about their detailed speech (song) airflow patterns, in addition to the existing audio, and chest/mid riff/belly movement (large-scale breathing patterns) capturing systems.
  • Another application is treatment for speech impediments due to one or more of traumatic brain injury, deaf speech, cleft lip/palate speech, or any other issues that influence oral airflow, nasal airflow, or nasality in speech.
  • Another application is telemedicine. Visualisations of oral and nasal production are transmitted to a therapist who is physically distant from the patient. The therapist can thus interact with a remote patient, simultaneously receiving audio, video, and oral/nasal flow information. This information enables various forms of therapy, that at present require the therapist to be present to feel the oral/nasal production, to be offered to patients who would otherwise have no access to such therapy (elderly, disabled, remote areas).
  • Bio-feedback includes showing the flow information, a processed form of the flow information (such as a nasalence measure), computed features (such as fO), or indirect visualisations and game objectives.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Pulmonology (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Physiology (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

Un système de mesure de flux d'air en situation de parole comprend un module d'extraction de caractéristiques, configuré pour recevoir des signaux d'entrée associés à un utilisateur à partir d'un premier et d'un second capteur, et pour estimer la forme et/ou la vitesse du flux d'air au moins en partie à partir des signaux d'entrée. Le système peut en outre comprendre un casque d'écoute comprenant le premier capteur positionné à l'intérieur au moins du premier flux d'air de l'utilisateur; le second capteur positionné à l'intérieur au moins du second flux d'air de l'utilisateur, et un élément de protection pour protéger le premier capteur du second flux d'air. L'élément de protection permet de former un dégagement entre l'élément de protection et le visage de l'utilisateur, lorsque le casque d'écoute est en cours d'utilisation.
PCT/NZ2017/050104 2016-07-27 2017-07-27 Système de mesure de flux d'air en situation de parole sans masque WO2018021920A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
NZ722578 2016-07-27
NZ72257816 2016-07-27

Publications (1)

Publication Number Publication Date
WO2018021920A1 true WO2018021920A1 (fr) 2018-02-01

Family

ID=61016311

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/NZ2017/050104 WO2018021920A1 (fr) 2016-07-27 2017-07-27 Système de mesure de flux d'air en situation de parole sans masque

Country Status (1)

Country Link
WO (1) WO2018021920A1 (fr)

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3637433A1 (fr) 2018-10-11 2020-04-15 Cordio Medical Ltd. Estimation du volume pulmonaire par analyse vocale
US10704021B2 (en) 2012-03-15 2020-07-07 Flodesign Sonics, Inc. Acoustic perfusion devices
US10724029B2 (en) 2012-03-15 2020-07-28 Flodesign Sonics, Inc. Acoustophoretic separation technology using multi-dimensional standing waves
US10785574B2 (en) 2017-12-14 2020-09-22 Flodesign Sonics, Inc. Acoustic transducer driver and controller
US10796805B2 (en) 2015-10-08 2020-10-06 Cordio Medical Ltd. Assessment of a pulmonary condition by speech analysis
US10967298B2 (en) 2012-03-15 2021-04-06 Flodesign Sonics, Inc. Driver and control for variable impedence load
US10975368B2 (en) 2014-01-08 2021-04-13 Flodesign Sonics, Inc. Acoustophoresis device with dual acoustophoretic chamber
US11007457B2 (en) 2012-03-15 2021-05-18 Flodesign Sonics, Inc. Electronic configuration and control for acoustic standing wave generation
US11011188B2 (en) 2019-03-12 2021-05-18 Cordio Medical Ltd. Diagnostic techniques based on speech-sample alignment
US11021699B2 (en) 2015-04-29 2021-06-01 FioDesign Sonics, Inc. Separation using angled acoustic waves
US11024327B2 (en) 2019-03-12 2021-06-01 Cordio Medical Ltd. Diagnostic techniques based on speech models
WO2021138648A1 (fr) * 2020-01-03 2021-07-08 Starkey Laboratories, Inc. Dispositif électronique porté sur l'oreille utilisant une adaptation d'environnement acoustique
US11085035B2 (en) 2016-05-03 2021-08-10 Flodesign Sonics, Inc. Therapeutic cell washing, concentration, and separation utilizing acoustophoresis
US11214789B2 (en) 2016-05-03 2022-01-04 Flodesign Sonics, Inc. Concentration and washing of particles with acoustics
US11377651B2 (en) 2016-10-19 2022-07-05 Flodesign Sonics, Inc. Cell therapy processes utilizing acoustophoresis
US11417342B2 (en) 2020-06-29 2022-08-16 Cordio Medical Ltd. Synthesizing patient-specific speech models
US11420136B2 (en) 2016-10-19 2022-08-23 Flodesign Sonics, Inc. Affinity cell extraction by acoustics
US11459540B2 (en) 2015-07-28 2022-10-04 Flodesign Sonics, Inc. Expanded bed affinity selection
US11474085B2 (en) 2015-07-28 2022-10-18 Flodesign Sonics, Inc. Expanded bed affinity selection
US11484211B2 (en) 2020-03-03 2022-11-01 Cordio Medical Ltd. Diagnosis of medical conditions using voice recordings and auscultation
US11708572B2 (en) 2015-04-29 2023-07-25 Flodesign Sonics, Inc. Acoustic cell separation techniques and processes
US12035107B2 (en) 2020-01-03 2024-07-09 Starkey Laboratories, Inc. Ear-worn electronic device employing user-initiated acoustic environment adaptation

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4519399A (en) * 1982-12-13 1985-05-28 Rion Kabushiki Kaisha Method for measuring the degree of nasality
US6850882B1 (en) * 2000-10-23 2005-02-01 Martin Rothenberg System for measuring velar function during speech
US20130090927A1 (en) * 2011-08-02 2013-04-11 Massachusetts Institute Of Technology Phonologically-based biomarkers for major depressive disorder
US20160081611A1 (en) * 2014-09-24 2016-03-24 International Business Machines Corporation System and method to measure, analyze, and model pulmonary function and disease utilizing temporal, spatial, and contextual data
WO2016128627A1 (fr) * 2015-10-14 2016-08-18 Aerophonoscope Appareil de mesure des respirations buccales et nasales d'un être humain

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4519399A (en) * 1982-12-13 1985-05-28 Rion Kabushiki Kaisha Method for measuring the degree of nasality
US6850882B1 (en) * 2000-10-23 2005-02-01 Martin Rothenberg System for measuring velar function during speech
US20130090927A1 (en) * 2011-08-02 2013-04-11 Massachusetts Institute Of Technology Phonologically-based biomarkers for major depressive disorder
US20160081611A1 (en) * 2014-09-24 2016-03-24 International Business Machines Corporation System and method to measure, analyze, and model pulmonary function and disease utilizing temporal, spatial, and contextual data
WO2016128627A1 (fr) * 2015-10-14 2016-08-18 Aerophonoscope Appareil de mesure des respirations buccales et nasales d'un être humain

Cited By (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10704021B2 (en) 2012-03-15 2020-07-07 Flodesign Sonics, Inc. Acoustic perfusion devices
US10724029B2 (en) 2012-03-15 2020-07-28 Flodesign Sonics, Inc. Acoustophoretic separation technology using multi-dimensional standing waves
US11007457B2 (en) 2012-03-15 2021-05-18 Flodesign Sonics, Inc. Electronic configuration and control for acoustic standing wave generation
US10967298B2 (en) 2012-03-15 2021-04-06 Flodesign Sonics, Inc. Driver and control for variable impedence load
US10975368B2 (en) 2014-01-08 2021-04-13 Flodesign Sonics, Inc. Acoustophoresis device with dual acoustophoretic chamber
US11708572B2 (en) 2015-04-29 2023-07-25 Flodesign Sonics, Inc. Acoustic cell separation techniques and processes
US11021699B2 (en) 2015-04-29 2021-06-01 FioDesign Sonics, Inc. Separation using angled acoustic waves
US11459540B2 (en) 2015-07-28 2022-10-04 Flodesign Sonics, Inc. Expanded bed affinity selection
US11474085B2 (en) 2015-07-28 2022-10-18 Flodesign Sonics, Inc. Expanded bed affinity selection
US10796805B2 (en) 2015-10-08 2020-10-06 Cordio Medical Ltd. Assessment of a pulmonary condition by speech analysis
US11085035B2 (en) 2016-05-03 2021-08-10 Flodesign Sonics, Inc. Therapeutic cell washing, concentration, and separation utilizing acoustophoresis
US11214789B2 (en) 2016-05-03 2022-01-04 Flodesign Sonics, Inc. Concentration and washing of particles with acoustics
US11420136B2 (en) 2016-10-19 2022-08-23 Flodesign Sonics, Inc. Affinity cell extraction by acoustics
US11377651B2 (en) 2016-10-19 2022-07-05 Flodesign Sonics, Inc. Cell therapy processes utilizing acoustophoresis
US10785574B2 (en) 2017-12-14 2020-09-22 Flodesign Sonics, Inc. Acoustic transducer driver and controller
EP3637433A1 (fr) 2018-10-11 2020-04-15 Cordio Medical Ltd. Estimation du volume pulmonaire par analyse vocale
US10847177B2 (en) 2018-10-11 2020-11-24 Cordio Medical Ltd. Estimating lung volume by speech analysis
US11024327B2 (en) 2019-03-12 2021-06-01 Cordio Medical Ltd. Diagnostic techniques based on speech models
US11011188B2 (en) 2019-03-12 2021-05-18 Cordio Medical Ltd. Diagnostic techniques based on speech-sample alignment
WO2021138648A1 (fr) * 2020-01-03 2021-07-08 Starkey Laboratories, Inc. Dispositif électronique porté sur l'oreille utilisant une adaptation d'environnement acoustique
US12069436B2 (en) 2020-01-03 2024-08-20 Starkey Laboratories, Inc. Ear-worn electronic device employing acoustic environment adaptation for muffled speech
US12035107B2 (en) 2020-01-03 2024-07-09 Starkey Laboratories, Inc. Ear-worn electronic device employing user-initiated acoustic environment adaptation
US11484211B2 (en) 2020-03-03 2022-11-01 Cordio Medical Ltd. Diagnosis of medical conditions using voice recordings and auscultation
US11417342B2 (en) 2020-06-29 2022-08-16 Cordio Medical Ltd. Synthesizing patient-specific speech models

Similar Documents

Publication Publication Date Title
WO2018021920A1 (fr) Système de mesure de flux d'air en situation de parole sans masque
Mehta et al. Relationships between vocal function measures derived from an acoustic microphone and a subglottal neck-surface accelerometer
Rothenberg Measurement of airflow in speech
Mehta et al. Mobile voice health monitoring using a wearable accelerometer sensor and a smartphone platform
Shadle The aerodynamics of speech
Moussavi Fundamentals of respiratory sounds and analysis
US20160302003A1 (en) Sensing non-speech body sounds
Nallanthighal et al. Deep learning architectures for estimating breathing signal and respiratory parameters from speech recordings
Mussell The need for standards in recording and analysing respiratory sounds
Nallanthighal et al. Speech breathing estimation using deep learning methods
JPH01183700A (ja) 口頭空気流を利用した音声バラメーターの抽出法とその装置
CN107898432B (zh) 鼾声检测方法、装置、设备及介质
Shadle et al. Fluid flow in a dynamic mechanical model of the vocal folds and tract. II. Implications for speech production studies
Cheyne Estimating glottal voicing source characteristics by measuring and modeling the acceleration of the skin on the neck
JP2014064675A (ja) 非拘束無呼吸検知システムとその方法とそのプログラム
Carullo et al. Performance comparison of different contact microphones used for voice monitoring
Meltzner et al. Measuring the neck frequency response function of laryngectomy patients: Implications for the design of electrolarynx devices
Derrick et al. Maskless speech airflow estimation system
Zhdanov et al. Short review of devices for detection of human breath sounds and heart tones
Curran et al. Using acoustic sensors to discriminate between nasal and mouth breathing
Akafi et al. Detection of hypernasal speech in children with cleft palate
Krecichwost et al. Multichannel speech acquisition and analysis for computer-aided sigmatism diagnosis in children
Nallanthighal et al. COVID-19 detection based on respiratory sensing from speech
Lee et al. Contributions of forward-focused voice to audio-vocal feedback measured using nasal accelerometry and power spectral analysis of vocal fundamental frequency
Netsell A perceptual-acoustic-physiological study of syllable stress

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17834846

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17834846

Country of ref document: EP

Kind code of ref document: A1