Low Bandwidth PERS Button Using Neural Network Audio Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Personal Emergency Response Systems (PERS) face challenges in providing effective verbal communication due to the need for full cellular communications channels, which are costly and complex, especially in mobile solutions where a 24/7 guaranteed channel is required, and the difficulty in maintaining suitable communications quality in a small help button format.
Innovation Solution
A personal help button and administrator system that utilize a trained neural network to convert verbal audio inputs into low bandwidth format signals, allowing for communication over low bandwidth channels such as Narrowband IoT, LoRa, or Sigfox, avoiding the need for conventional cellular networks, and enabling efficient data compression and speech synthesis for verbal audio delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full cellular communications channel is used for verbal communication between user and PERS agent, then communication quality is improved, but device cost and system complexity increase
Solution Approach 1:
The patent extracts only the essential information from verbal communication (keywords, phrases, intent) rather than transmitting the full audio signal. The neural network processes audio locally at the help button device, extracting semantic content and transmitting only this compressed representation over low-bandwidth cellular channels, thereby maintaining communication effectiveness while eliminating the need for high-bandwidth audio transmission infrastructure
Solution Approach 2:
The patent fundamentally changes the parameter of communication bandwidth from high (full audio quality requiring cellular-grade channels) to low (textual/semantic representation). By transforming audio signals into compressed semantic representations through neural network processing, the system operates on low-bandwidth channels that are significantly simpler and less costly than traditional cellular voice channels
2Reliability
If full cellular communications channel is used for verbal communication, then communication quality is improved, but equipment cost and rental cost increase
Solution Approach 1:
The patent extracts only the essential information from verbal communication (keywords, phrases, intent) rather than transmitting the full audio signal. The neural network processes audio locally at the help button device, extracting semantic content and transmitting only this compressed representation over low-bandwidth cellular channels, thereby maintaining communication effectiveness while eliminating the need for high-bandwidth audio transmission infrastructure
Solution Approach 2:
The patent replaces expensive, resource-intensive full cellular voice communication infrastructure with a much cheaper alternative: low-bandwidth text/semantic transmission. The system uses minimal cellular data channels that are significantly less costly in terms of equipment, bandwidth rental, and operational expenses, while the neural network processing occurs locally at the disposable help button device
3Reliability
If large bandwidth audio signals are used for verbal communication, then communication quality is improved, but range of the signal is reduced
Solution Approach 1:
The patent fundamentally changes the parameter of communication bandwidth from high (full audio quality requiring cellular-grade channels) to low (textual/semantic representation). By transforming audio signals into compressed semantic representations through neural network processing, the system operates on low-bandwidth channels that have extended range and better penetration characteristics compared to high-bandwidth audio signals
4Reliability
If conventional cellular networks are used for communication, then communication quality is improved, but device complexity and power consumption increase
Solution Approach 1:
The patent extracts only the essential information from verbal communication (keywords, phrases, intent) rather than transmitting the full audio signal. The neural network processes audio locally at the help button device, extracting semantic content and transmitting only this compressed representation over low-bandwidth cellular channels, thereby maintaining communication effectiveness while eliminating the need for high-bandwidth audio transmission infrastructure
Solution Approach 2:
The patent fundamentally changes the parameter of communication bandwidth from high (full audio quality requiring cellular-grade channels) to low (textual/semantic representation). By transforming audio signals into compressed semantic representations through neural network processing, the system operates on low-bandwidth channels that are significantly simpler and less costly than traditional cellular voice channels
Data Source
AI summary
A personal help button and the administrator part of a personal emergency response system are provided, in which verbal audio is converted into a low bandwidth format signal. A conversation algorithm is used which comprises a trained neural network. The low bandwidth format has a bandwidth below 500 bits/second.


