Transient Noise Rejection in Speech Recognition Pre-processor
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automatic speech recognition systems face challenges in accurately recognizing speech due to the presence of transient noise, such as ambient and non-speech vocalizations, which can lead to rejection errors, especially in environments like vehicles where speakers of tonal languages face significant noise interference.
Innovation Solution
A method that involves receiving audio with user speech and transient noise, converting it into digital data, segmenting it into acoustic frames, evaluating each frame for transient noise, rejecting noisy frames, and accepting clean frames for speech recognition, thereby improving the accuracy of speech recognition systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If acoustic frames are processed for speech recognition without filtering, then processing speed is maintained, but recognition accuracy deteriorates due to transient noise
Solution Approach 1:
The patent applies preliminary action by evaluating acoustic frames for transient noise before they are processed by the speech recognition decoder. The pre-processor identifies and rejects noisy frames in advance, preventing them from degrading recognition accuracy. This early filtering approach maintains processing efficiency while improving accuracy by only allowing clean frames to proceed through the full recognition pipeline.
2Reliability
If all acoustic frames are accepted for recognition, then no data is lost, but recognition reliability worsens due to transient noise interference
Solution Approach 1:
The patent applies the extraction principle by separating and removing only the harmful transient noise frames from the acoustic stream, while preserving all clean speech frames. The pre-processor evaluates each frame and selectively rejects only those containing transient noise, rather than discarding entire speech segments. This targeted removal maintains speech data integrity while eliminating noise interference.
3Measurement precision
If transient noise filtering is applied frame-by-frame, then recognition accuracy improves, but system complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the continuous acoustic stream into discrete frames and evaluating each frame independently for transient noise. This frame-by-frame approach allows the system to apply simple noise detection criteria to each segment, avoiding the need for complex global analysis. The segmented processing makes the filtering operation computationally efficient while maintaining high recognition accuracy.
Data Source
AI summary
A method of and system for transient noise rejection for improved speech recognition. The method comprises the steps of (a) receiving audio including user speech and at least some transient noise associated with the speech, (b) converting the received audio into digital data, (c) segmenting the digital data into acoustic frames, and (d) extracting acoustic feature vectors from the acoustic frames. The method also comprises the steps of (e) evaluating the acoustic frames for transient noise on a frame-by-frame basis, (f) rejecting those acoustic frames having transient noise, (g) accepting as speech frames those acoustic frames having no transient noise and, thereafter, (h) recognizing the user speech using the speech frames.