Prediction and identification techniques used with a hearing prosthesis
Inventors
Carter, Paul Michael • Hersbach, Adam • Murphy, Richard Bruce • OPLINGER, Kenneth
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12236942-B2
Publication Date
2025-02-25
Expiration Date
Abstract
A method, including receiving a signal which includes speech data, processing the received signal to identify and/or predict one or more words in the speech data, and evoking a hearing percept based on the received signal, wherein the evoked hearing percept includes one or more modified words based on the identification and/or prediction of the one or more words.
Core Innovation
The invention relates to hearing prosthesis systems that receive a signal including speech data, process the received signal to identify and/or predict one or more words in the speech data, and evoke a hearing percept based on the received signal. The evoked hearing percept includes one or more modified words based on the identification and/or prediction of the one or more words, so that the hearing percept is different relative to that which would have been the case in the absence of the identification and/or prediction.
A key aspect is that the processing introduces a delay between receiving the signal and evoking the hearing percept, and the processing may be configured so it is not correlated with a noticeable delay in the device used to execute the evoking of the hearing percept noticed by the person in which the hearing percept is evoked. The invention also covers modifying an existing signal or producing a new signal so that the resulting hearing percept includes modified words, including a totally synthesized word replacement and speech that is intentionally different.
The invention further includes hearing percept generation that leverages prediction to influence hearing outcomes, including automatically adjusting a noise control system and automatically adjusting a volume and/or a gain of a hearing prosthesis based on the identified and/or predicted one or more words. In addition, the invention includes a speech presence probability (SPP) estimator enhanced using word prediction, and supports using sound prediction functionality to predict sound captured by a sound capture device.
In an example system architecture, a device includes a sound capture device and a processing system that converts sound captured by the system and process a signal based on the captured sound with the processing system to evoke a hearing percept based on output from the processing system. The system includes sound prediction functionality and may include word prediction functionality, prediction-based operation of noise reduction, enhancement of a speech presence probability estimator, and delay behavior including environment/context-dependent delay relative to a normal processing delay, with implementation potentially distributed across a hearing prosthesis and a handheld or geographically remote server.
Claims Coverage
The document includes three independent claims (clm-00001, clm-00007, clm-00009) and a system claim (clm-00017). Across these independent claims, the inventive features center on identifying and/or predicting words from speech data to evoke a hearing percept with modified words, with optional delay behavior, and with prosthesis-adaptive features such as noise control/volume-gain adjustment, hybrid/totally synthesized percept generation, and certainty-level or speech presence probability estimator enhancements.
Identify and/or predict words from speech data to modify an evoked hearing percept
A method receives a signal which includes speech data, processes the received signal to identify and/or predict one or more words in the speech data, and evokes a hearing percept based in the received signal, wherein the evoked hearing percept includes one or more modified words based on the identification and/or prediction of the one or more words.
Introduce a delay between receiving the signal and evoking the hearing percept
The method introduces a delay between receiving the signal and evoking the hearing percept.
Evoked hearing percept not correlated with a noticeable delay in the executing device
The method includes processing such that it is not correlated with a noticeable delay in a device used to execute the evoking of the hearing percept noticed by the person in which the hearing percept is evoked.
Evoked hearing percept different relative to absence of identification and/or prediction
The method evokes a hearing percept based in the received signal, wherein the evoked hearing percept is a different hearing percept relative to that which would have been the case in the absence of the identification and/or prediction.
Automatically adjust a noise control system of a hearing prosthesis based on identified and/or predicted words
The method includes automatically adjusting a noise control system of a hearing prosthesis that is evoking the hearing percept based on the identified and/or predicted one or more words.
Automatically adjust a volume and/or a gain of a hearing prosthesis based on identified and/or predicted words
The method includes automatically adjusting a volume and/or a gain of a hearing prosthesis that is evoking the hearing percept based on the identified and/or predicted one or more words.
Provide a hybrid word based on identification results
The action of evoking the hearing percept is executed using a hearing prosthesis that is configured to evoke the hearing percept such that a hybrid word is provided to the recipient based on the results of the identification.
Evaluate a level of certainty and generate a certainty-dependent modified hearing percept
The action of evoking the hearing percept is executed using a hearing prosthesis that is configured to evaluate a level of certainty of the identification and generate a hearing percept based on the evaluation of the level of certainty, wherein the generated hearing percept is different depending on the level of certainty.
Totally synthesized hearing percept optimized by tradeoff
The evoked hearing percept is a totally synthesized hearing percept optimized to maximize speech comprehension while trading off tone, uniqueness and/or emphasis.
Sound capture and processing with sound prediction to predict captured sound
A system includes a device including a sound capture device and a processing system, wherein the device is configured to convert sound captured by the system and process a signal based on the captured sound with the processing system and evoke a hearing percept based on output from the processing system based on the signal, wherein the system includes a sound prediction and/or word identification functionality such that the system is configured to predict sound that is captured by the sound capture device.
Across the independent claims, the core claim coverage is for methods and systems that predict or identify one or more words in speech data and evoke a hearing percept containing modified words, with additional selectable claim-scoped capabilities including delay handling, non-correlated timing, and hearing-prosthesis adaptations such as noise control adjustment and volume/gain adjustment, plus hybrid or totally synthesized percept generation and certainty-dependent percept generation. The system independent claim additionally covers sound prediction functionality configured to predict sound captured by the sound capture device.
Stated Advantages
The evoked hearing percept includes one or more modified words based on identification and/or prediction.
The evoked hearing percept is different relative to that which would have been the case in the absence of the identification and/or prediction.
Automatically adjusting a noise control system of a hearing prosthesis based on the identified and/or predicted one or more words.
Automatically adjusting a volume and/or a gain of a hearing prosthesis based on the identified and/or predicted one or more words.
A hearing prosthesis can provide a hybrid word based on the results of the identification.
A hearing prosthesis can evaluate a level of certainty of the identification and generate a hearing percept different depending on the level of certainty.
A totally synthesized hearing percept is optimized to maximize speech comprehension while trading off tone, uniqueness and/or emphasis.
A speech presence probability (SPP) estimator is enhanced by using sound prediction functionality.
Sound prediction functionality is configured to predict sound that is captured by the sound capture device.
Documented Applications
Hearing prosthesis use where a processed speech signal with word identification/prediction evokes a modified hearing percept for a recipient.
Noise control and speech enhancement in a hearing prosthesis by automatically adjusting a noise control system based on identified and/or predicted words.
Systems that include sound prediction to predict sound captured by a sound capture device in connection with evoking a hearing percept.
Speech presence probability (SPP) estimation enhanced using word prediction in the context of a hearing prosthesis system.
Delay and timing behavior in a hearing prosthesis where processing introduces a delay between receiving the signal and evoking the hearing percept, including context/environment-dependent delay relative to a normal processing delay.
Distributed implementations that may include a handheld remote device and/or a geographically remote server as part of a hearing prosthesis system architecture.
Interested in licensing this patent?