Real-Time Speech Recognition and Native Visualization via Split Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data analytics systems lack real-time speech recognition and native visualization tools, limiting user interaction and insights generation during data analysis processes.
Innovation Solution
A system and method for providing real-time speech recognition and native visualization, enabling on-device speech-to-text recognition, local processing of sentences, and server-based intent determination for immediate data analytics results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If speech recognition and data processing are performed entirely on the server, then processing capability and accuracy are improved, but response time increases and real-time interaction is degraded
Solution Approach 1:
The system divides speech processing into two segments: on-device speech-to-text conversion using built-in engines, and server-based intent determination and data analysis. This segmentation allows rapid local processing of speech signals while transferring only the converted text to the server, reducing response time while maintaining accuracy.
Solution Approach 2:
The system performs preliminary speech-to-text conversion on the device before sending data to the server. By pre-converting speech to text locally using built-in recognition engines, the system eliminates the need to transmit audio data to the server, reducing network dependency and response time while preserving speech recognition accuracy.
2Loss of time
If speech processing is performed entirely on the device, then response time is reduced and real-time interaction is improved, but processing capability and complexity of handling diverse intents are limited
Solution Approach 1:
The system segments the processing tasks by assigning speech-to-text conversion to the device (leveraging built-in engines) and intent determination to the server (leveraging advanced NLP capabilities). This division allows the device to respond quickly while the server handles complex analytical tasks.
Solution Approach 2:
The system uses an intermediary communication protocol between device and server that transmits only essential text data and receives structured intent results. This intermediary layer enables the device to maintain real-time responsiveness while the server provides sophisticated intent analysis capabilities.
3Productivity
If all data processing is performed on the server, then processing power is sufficient, but network dependency increases and offline functionality is lost
Solution Approach 1:
The system performs preliminary speech-to-text conversion on the device before any server communication is needed. This preliminary local processing ensures that basic functionality remains available offline, while server connectivity enables enhanced capabilities when available.
4Measurement precision
If speech-to-text conversion is performed on the server, then recognition accuracy is improved, but network bandwidth consumption increases and response time is delayed
Solution Approach 1:
The system performs preliminary speech-to-text conversion on the device using built-in recognition engines before transmitting data to the server. This eliminates the need to upload audio files, significantly reducing network bandwidth consumption and response time while maintaining adequate recognition accuracy for analytics purposes.
Data Source
AI summary
Embodiments described herein are generally related to computer data analytics, and computer-based methods of providing business intelligence or other data, and are particularly related to a system and method for providing a real-time speech recognition and native visualization tool, for use with data analytics environments. The described approach can be used to provide real-time data visualizations or other analytics based on a live conversation with a user using an on-device speech-to-text recognition engine. In accordance with an embodiment, the method includes: listening for audio received as input from a user; building an array of words associated with a sentence; determining whether the sentence is capable of being processed without additional server data and if so performing the step locally; and alternatively sending an input sentence to a server API, in order to determine an intent associated therewith for use by the server in returning relevant results.