Asynchronous Speech Recognition Offline Audio Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing speech-to-text systems that rely on remote hosted applications face issues with audio quality due to retransmission over public networks, intermittent connectivity, and restricted access in secure environments, leading to unavailability of SaaS processing when connectivity is lost.

Innovation Solution

The implementation of an asynchronous speech recognition method that allows a client device to record audio and generate context files locally when connectivity is unavailable, enabling offline processing and synchronization with the hosted application when connectivity is restored.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If speech-to-text processing is performed remotely via hosted application, then processing power and accuracy are improved, but system availability deteriorates when connectivity is lost

Engineering Contradiction:
Improvespeech-to-text processing accuracyVSAvoidsystem availability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary actions by recording audio and generating context files locally before connectivity is available. The client device captures audio data, creates context files with metadata, and stores them in a queue for later processing, enabling the system to be ready for immediate processing once connectivity is restored.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary layer consisting of local audio recording, context file generation, and a queue mechanism. This intermediary allows the system to decouple the audio capture function from the remote processing function, enabling offline operation while maintaining the ability to leverage remote processing power when available.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If audio is retransmitted over public networks, then remote processing is enabled, but audio quality deteriorates

Engineering Contradiction:
Improveremote processing capabilityVSAvoidaudio quality
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system records audio locally in its original high-quality form before any network transmission occurs. By capturing and storing the raw audio data locally first, the system preserves audio fidelity and only transmits necessary metadata (context files) over the network, avoiding quality degradation from repeated retransmissions.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If speech-to-text application is co-resident with local computer, then system availability is improved, but processing power is limited

Engineering Contradiction:
Improvesystem availabilityVSAvoidprocessing capability
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The system segments the speech-to-text functionality into distinct components: local audio recording, local context file generation, remote speech-to-text processing, and local result integration. This segmentation allows each component to operate independently at its optimal location, combining local availability with remote processing power.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system transitions from a single-dimension architecture (either local or remote processing) to a multi-dimensional hybrid architecture. It operates across spatial dimensions (local device and remote server) and temporal dimensions (synchronous and asynchronous operation), allowing flexible deployment based on connectivity and processing needs.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

4Measurement precision

If hosted application is used in secure environments, then processing power is improved, but access is restricted when connectivity is limited

Engineering Contradiction:
Improveprocessing powerVSAvoidenvironmental adaptability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system dynamically adapts its operation mode based on environmental conditions. When connectivity is available, it leverages remote hosted application processing. When connectivity is restricted or unavailable, it automatically switches to local recording and queue-based asynchronous processing, making the system versatile across different environmental constraints.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3966805B1Asynchronous speech to text data processing
Publication Date: 2025.03.05 NVOQ INC
  • EP3966805B1 patent drawingFigure 1
  • EP3966805B1 patent drawingFigure 2
  • EP3966805B1 patent drawingFigure 3

AI summary

A method to allow for asynchronous speech recognition for a primary application's use is provided. The method comprises evoking a primary application and a client device APP to work with a remote hosted application to process audio for the primary application. The APP connects to the hosted application, and if successful, the processing proceeds. If the APP cannot connect to the hosted application, the APP generates an input data file and a context file. The input data file may be an audio file in certain embodiments to record audio of a user dictating to the client device's microphone. The context file contains, among other things, the application information and navigation information such that the audio, once processed, may be inserted to the primary application based on the data contained in the context file.