Voice Recognition Signal Processing for Interference Cancellation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice recognition methods have a low recognition success rate due to significant differences between the voice signal to be recognized and the pure voice signal, often contaminated with interfering sounds like echo and noise.
Innovation Solution
A voice recognition method that performs high-pass filtering, cancels interfering sounds using adaptive filtering and noise suppression algorithms, and applies automatic gain control to obtain a target voice signal, which is then processed to extract a feature vector for input into a pre-trained acoustic model for improved recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If voice recognition is performed using a pure voice signal acoustic model, then the acoustic model can be trained with clean data, but the recognition success rate is low when processing real voice signals containing interfering sounds
Solution Approach 1:
The patent applies preliminary signal processing actions (high-pass filtering, echo cancellation, noise suppression, automatic gain control) to the voice signal before recognition. This prepares the signal in advance to match the acoustic model's expectations, resolving the mismatch between clean training data and noisy real-world signals.
Solution Approach 2:
The patent transforms the voice signal through multiple parameter changes: frequency domain filtering (high-pass), temporal processing (echo cancellation), spectral processing (noise suppression), and amplitude normalization (AGC). These parameter transformations convert the raw signal into a form that matches the acoustic model's training conditions.
2Measurement precision
If signal processing operations are applied to remove interfering sounds, then recognition accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent segments the signal processing task into four distinct modules: high-pass filtering, echo cancellation, noise suppression, and automatic gain control. Each module handles a specific aspect of interference removal, making the overall complex process manageable and systematic.
Solution Approach 2:
The patent introduces intermediate processing stages between the raw voice signal and the acoustic model. Each processing stage acts as an intermediary that progressively cleans and prepares the signal, with the final processed signal serving as the intermediary input to the recognition model.
Data Source
AI summary
The present application discloses a voice recognition method and apparatus. A specific implementation of the method includes: in response to detecting a microphone receiving voice signal containing interfering sound signal, performing high-pass filtering on the voice signal; cancelling the interfering sound signal in the voice signal subjected to high-pass filtering; performing automatic gain control on the voice signal subjected to cancelling the interfering sound signal, to obtain target voice signal; and extracting a feature vector from the target voice signal and inputting the feature vector into a pre-trained acoustic model, to obtain a voice recognition result matching the target voice signal, the acoustic model being used for representing a corresponding relationship between the feature vector and the voice recognition result. This implementation improves the success rate of voice recognition.