Hearing Aid Speech Intelligibility via Remote Text-to-Speech Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern digital hearing aids face challenges in improving speech intelligibility, especially in noisy environments and with idiosyncratic speech patterns, due to the complexity of speech recognition and processing power limitations, particularly when handling foreign languages or filler words.
Innovation Solution
A system that utilizes a remote server with a Speech Recognition Engine and an external device with a Text-To-Speech engine to process audio streams, converting speech to text and synthesizing speech based on text, thereby removing idiosyncratic speech artifacts, and compensating for hearing loss through frequency-dependent gain and linear frequency transposition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speech recognition is performed locally in the hearing aid, then speech intelligibility can be improved, but the processing power requirements exceed the capabilities of hearing aid devices
Solution Approach 1:
The patent introduces an external device (smartphone, tablet, or server) as an intermediary between the hearing aid and the speech recognition processing. The hearing aid captures audio and transmits it to the external device which performs the computationally intensive speech recognition and text-to-speech conversion, then returns the processed audio to the hearing aid. This mediator approach allows complex processing to be performed outside the hearing aid while maintaining the hearing aid's portability and low power consumption characteristics.
Solution Approach 2:
The patent moves the speech recognition processing from the spatial dimension (local processing within the hearing aid) to the network dimension (remote processing via Internet connection). By utilizing cloud-based servers accessible through Internet protocols, the system offloads processing requirements to remote infrastructure, enabling sophisticated speech processing without constraining the hearing aid's hardware capabilities.
2Reliability
If sophisticated speech recognition processing is implemented, then speech intelligibility in noisy environments improves, but the device complexity and processing requirements increase significantly
Solution Approach 1:
The external device serves as an intermediary that handles all complex speech recognition, noise suppression, and text-to-speech processing. The hearing aid itself remains relatively simple, functioning primarily as an audio capture and playback device with Bluetooth connectivity. This division of labor reduces the complexity burden on the hearing aid while achieving sophisticated speech processing results.
Solution Approach 2:
The system creates a digital copy of the audio signal and processes this copy through sophisticated algorithms on the external device, while the original audio path remains simple. The processed speech is then synthesized and returned to the hearing aid for playback, allowing complex processing without permanently complicating the hearing aid's architecture.
3Loss of time
If real-time speech processing is performed locally, then latency is reduced, but the hearing aid lacks sufficient processing power for accurate speech recognition
Solution Approach 1:
The system performs preliminary actions by pre-processing audio data during brief intervals when the hearing aid is not actively playing audio, and by pre-establishing Bluetooth connections and buffering data. The external device also maintains ready-state processing capabilities, allowing rapid response when audio processing is requested, thus reducing effective latency despite the distributed architecture.
Data Source
AI summary
A system for improving speech intelligibility of an audio signal, comprises a hearing aid (10), a server (71), and an external device (50)—such as a smartphone or a tablet computer. The hearing aid has a short range transceiver. The server is accessible via the Internet, and has a Speech Recognition Engine converting speech into text. The external device has a short range transceiver for communication with said hearing aid, a second transceiver for providing a wireless data connection to said server via the Internet, means for handling a speech stream intended for the hearing aid, and a Text-To-Speech engine adapted to synthesize speech based on a string of text. The invention also provides a method of improving speech intelligibility of an audio signal.