What is Software-Defined Hearing？
Software-Defined Hearing (SDH) is a groundbreaking technological framework that revolutionizes the hearing aid industry by separating the hearing aid functions from the hardware and abstracting them into a dedicated software layer. This approach allows for the implementation of core hearing capabilities, such as Wide Dynamic Range Compression (WDRC), noise reduction, and feedback management, through programmable code rather than hard-coding them into the silicon of a chip or an integrated circuit (IC). Drawing parallels with the evolution from feature phones to smartphones, SDH enables a single hardware platform to be dynamically updated and customized, providing users with continuous improvements in sound quality and personalized features without the need to replace the physical device.
The Architecture of Software-Defined Hearing
Lyratone's SDH architecture is structured into four distinct layers, facilitating a "Hearing as a Service" (HaaS) model that directly connects the user's ear to cloud intelligence:
1. Hardware Layer (System-on-Chip)
The base of the system is a high-performance System-on-Chip (SoC) that uses advanced process nodes (e.g., 12nm/14nm). This ensures the necessary computational power to run complex digital signal processing (DSP) and AI models with ultra-low latency (<15ms).
2. Embedded Application Layer
This layer houses the core hearing algorithms, including:
- WDRC (32–128 channels): For precise frequency compensation.
- AI Noise Reduction: Utilizing Convolutional Neural Networks (CNN) and Deep Neural Networks (DNN) models to isolate speech in noisy environments.
- Feedback Management: Real-time suppression of whistling (howling).
- Edge AI: Real-time inference performed directly on the device for immediate response.
3. Internet Application Layer
Serving as the communication gateway between the device and the user, this layer manages:
- Smart App Control: Allowing users to adjust parameters via a smartphone.
- Interactive Fitting: Automated hearing tests and parameter tuning based on user feedback.
- Connectivity Logic: Managing multi-link connections for telephony and streaming.
4. Cloud Layer
The "brain" of the platform, where Big Data and AI training occur, includes:
- Continuous Optimization: Analyzing anonymized user data to refine hearing prescriptions.
- AI Model Training: Developing more sophisticated deep learning models that are later pushed to the device via Over the Air (OTA) updates.
- Hearing as a Service: Enabling remote diagnostics, online consultations, and storage of personalized hearing profiles.
By leveraging high-performance, general-purpose SoC platforms, SDH creates an agile environment where hearing aids can evolve through software updates, offering users a seamless and continuously improving hearing experience.