Multi-Carrier Auditory Prosthetic Processing for Speech Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cochlear implants struggle to effectively deliver temporal fine structure cues, leading to difficulties in speech recognition in noise environments, as they often suppress all but one sound source or convey all sounds on a single carrier, resulting in poor speech intelligibility and limited TFS information transmission.
Innovation Solution
A multi-carrier processing method that dynamically adjusts the amplitude and frequency of separate carriers based on identified audio signals, allowing for the delivery of temporal fine structure information to cochlear implants, enabling users to naturally differentiate between target and background signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional cochlear-implant processors suppress all but one sound source to allow effective processing, then speech intelligibility of the target signal is improved, but the user loses the ability to switch between different sound sources and the system requires accurate target signal identification
Solution Approach 1:
The patent divides the audio signal into multiple frequency channels, each processed independently with its own carrier. This segmentation allows different sound sources to be distributed across multiple channels, enabling the user to access multiple sources simultaneously rather than suppressing all but one.
Solution Approach 2:
The patent introduces a new dimension by using multiple carriers with different pulse rates instead of a single carrier. This multi-carrier approach creates additional degrees of freedom in the signal representation, allowing the auditory system to segregate sources based on temporal fine structure differences across carriers.
2Device complexity
If all sounds are conveyed on a single carrier, then the system complexity is reduced, but speech intelligibility deteriorates significantly
Solution Approach 1:
Instead of using a single carrier for all sounds, the patent segments the audio signal into multiple frequency channels, each modulating its own carrier. This segmentation preserves speech intelligibility by maintaining spectral information while using multiple carriers rather than one.
Solution Approach 2:
The patent employs dynamic carrier pulse rates that can be adjusted based on the signal characteristics in each channel. This dynamic approach optimizes the representation of temporal fine structure for different sound sources and frequency regions, improving speech intelligibility compared to a fixed single-carrier system.
3Device complexity
If temporal envelope information is extracted and transmitted while discarding TFS information, then the processing is simplified, but the ability to extract audio signals from mixtures deteriorates
Solution Approach 1:
Instead of discarding TFS information as conventional systems do, the patent inverts the approach by preserving and transmitting TFS information through multiple carriers with different pulse rates. This inversion maintains the capability for source segregation while managing processing complexity through efficient multi-carrier modulation.
Solution Approach 2:
The patent changes the parameter of carrier pulse rate across different carriers to encode TFS information. By varying the pulse rate parameter of each carrier based on the TFS of the modulated signal, the system preserves source segregation cues without requiring complex processing of the original TFS waveform.
4Measurement precision
If the original TFS from the target speech is provided to cochlear-implant users, then speech recognition in noise is improved, but the technical implementation becomes challenging and fine structure transmission remains limited
Solution Approach 1:
The patent segments the target speech TFS information across multiple frequency channels, each modulating its own carrier. This segmentation makes the implementation of TFS transmission more manageable by breaking down the complex task of preserving original TFS into channel-specific processing operations.
Solution Approach 2:
The patent uses dynamic carrier pulse rates that adapt to the TFS characteristics of each channel. This dynamic modulation approach provides effective TFS transmission and improves speech recognition in noise while keeping the implementation feasible through efficient use of the multi-carrier framework.
Data Source
AI summary
A method for multi-carrier processing in an auditory prosthetic device comprises receiving an audio signal comprising multiple signals, each produced by an independent source, and identifying at least a first signal and a second signal in the received audio signal. The method also comprises adjusting an amplitude of a first carrier signal based on a first signal or a signal envelope associated with the first signal, and adjusting an amplitude of a second carrier signal based on a second signal or a signal envelope associated with the second signal. The amplitude-adjusted first and second signals are delivered to one or more stimulation devices associated with an auditory prosthetic.