Progressive Audio Encoding for Low Latency Speech Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Voice recognition applications on mobile devices face delays due to fluctuating mobile communication network quality, which affects the speed and fidelity of audio data transmission, leading to increased latency in receiving search results.
Innovation Solution
The method involves encoding audio input streams into sequential data blocks with progressively increasing data size, allowing the speech recognition server to reconstruct the audio with progressively higher fidelity by receiving and processing these blocks until a confident recognition is achieved, thereby reducing the need for transmitting the entire audio stream and improving latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire audio input stream is transmitted to the speech recognition server, then the recognition accuracy is improved, but the transmission time and latency are increased
Solution Approach 1:
The audio input stream is divided into multiple audio blocks, where each block contains a portion of the audio data. The mobile device transmits these blocks sequentially to the speech recognition server, allowing the server to begin processing earlier blocks while later blocks are still being transmitted, thereby reducing overall latency while maintaining recognition accuracy through cumulative processing of all blocks
Solution Approach 2:
The speech recognition server begins processing the audio data as soon as the first audio block is received, performing preliminary speech recognition attempts on incomplete data. This preliminary action allows the server to start working before receiving the complete audio stream, reducing the total time from when the user speaks to when recognition results are provided
2Speed
If audio data is transmitted in smaller blocks to reduce latency, then the transmission speed is improved, but the data completeness and recognition fidelity are reduced
Solution Approach 1:
The system maintains continuous processing of audio blocks as they arrive at the server. Each audio block is processed immediately upon receipt, and the speech recognition server continuously accumulates and processes subsequent blocks without interruption, ensuring that all audio information is eventually utilized for recognition while maintaining high transmission speed through the block-based approach
Data Source
AI summary
The present disclosure includes processing a signal to generate a first sub-set of data, transmitting the first sub-set of data for generation of a reconstructed audio signal, the reconstructed audio signal having a fidelity relative to the signal, processing the signal to generate a second sub-set of data and a third sub-set of data, the second sub-set of data defining a second portion of the signal and comprising data that is different than data of the first sub-set of data, and the third sub-set of data defining a third portion of the signal and comprising data that is different than data of the first and second sub-sets of data, comparing a priority of the second sub-set of data to a priority of the third sub-set of data, and transmitting one of the second sub-set of data and the third sub-set of data over the network for improving the fidelity.