Shared Voice Identification Models for Lower Network Traffic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Excessive network transmissions of network traffic data can lead to inefficient bandwidth utilization and complicate data routing, particularly in voice-activated systems, due to the computational intensity of audio identification models and the need for repeated training and transmission of audio samples across multiple devices.
Innovation Solution
A data processing system that enables sharing and cooperative access of audio identification models among client computing devices, reducing redundant computations and network transmissions by generating models once and updating them with audio samples from different locations, thereby improving accuracy and conserving resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If audio identification models are trained and transmitted repeatedly across multiple client devices, then each device can perform local audio identification, but network bandwidth is wasted and processing capacity is overloaded
Solution Approach 1:
A centralized server acts as an intermediary between client devices and audio identification models. The server receives audio samples from multiple clients, trains the audio identification model centrally, and distributes the trained model to clients. This eliminates the need for each client to independently train models, reducing redundant network transmissions and optimizing bandwidth utilization while maintaining identification accuracy.
Solution Approach 2:
The patent merges the audio identification model training function into a centralized server rather than distributing it across multiple client devices. By combining the training operations into a single location, the system eliminates redundant computations and network transmissions that would occur if each device trained its own model, thereby reducing network bandwidth consumption while preserving the ability to perform accurate local identification.
2Ease of operation
If each client device maintains its own audio identification model, then local processing is enabled, but computational resources are wasted through redundant training
Solution Approach 1:
The system performs preliminary model training action at the centralized server before distribution to client devices. The server receives audio samples from multiple clients, trains the audio identification model using aggregated data from all clients, and then distributes the pre-trained model to each client. This preliminary centralized training eliminates the need for each client to perform redundant training computations, improving overall computational efficiency while maintaining local processing capability.
Solution Approach 2:
Instead of each client device independently creating its own audio identification model through training, the system creates a single master model at the server and then copies this trained model to multiple client devices. This copying approach allows each client to have local processing capability with its own model instance, while avoiding the computational waste of redundant training operations at each device.
3Measurement precision
If audio samples are transmitted repeatedly for model training, then model accuracy can be improved, but network traffic increases and routing becomes complex
Solution Approach 1:
The centralized server acts as an intermediary that consolidates audio sample collection and model training operations. Instead of multiple clients independently transmitting audio samples to multiple destinations, all audio samples are routed through the central server, which manages the training process. This intermediary approach simplifies data routing complexity while still enabling improved model accuracy through aggregated training data from multiple clients.
Data Source
AI summary
The present disclosure is generally directed to a data processing system for customizing content in a voice activated computer network environment. With user consent, the data processing system can improve the efficiency and effectiveness of auditory data packet transmission over one or more computer networks by, for example, increasing the accuracy of the voice identification process used in the generation of customized content. The present solution can make accurate identifications while generating fewer audio identification models, which are computationally intensive to generate.