Speech Interface Audio Interference Cancellation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Devices interacting with users through speech face difficulties in isolating user speech in environments with interfering sounds from media sources like TVs, radios, or music players, which hinders accurate speech recognition and understanding.
Innovation Solution
A speech interface system that identifies interfering audio content by querying media players or online services, generates a signature or fingerprint of the interfering sound, and uses this information to suppress the noise through an interference cancellation process, allowing for more accurate automatic speech recognition and natural language understanding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If media players (TVs, radios, music players) are used in the user environment, then entertainment and media consumption are enhanced, but interfering sounds are generated that degrade speech recognition accuracy
Solution Approach 1:
The system obtains reference audio content from the same media sources that generate interfering sounds, then uses this reference content through adaptive filtering to cancel the interference. The harmful interfering sound is converted into a useful reference signal that enables its own cancellation, allowing speech recognition to proceed accurately despite the presence of media playback.
2Ease of operation
If speech recognition is performed in environments with media playback, then user interaction convenience is maintained, but speech isolation and recognition accuracy deteriorate due to interfering sounds
Solution Approach 1:
The system introduces reference audio content as an intermediary element between the media player and the speech recognition process. This reference content acts as a mediator that carries information about the interfering sound, enabling the adaptive filter to subtract the interference from the microphone input and isolate the user's speech accurately.
3Measurement precision
If interfering audio content is identified and suppressed using reference content from multiple sources, then speech recognition accuracy is improved, but system complexity increases
Solution Approach 1:
The system is designed to work with multiple types of media sources (TVs, radios, music players, streaming services) and content formats through a unified adaptive filtering framework. The same core algorithm handles different interference sources by obtaining reference content from various sources, reducing the need for source-specific processing logic and managing complexity through generalization.
Data Source
AI summary
A speech interface device may capture user speech for analysis by automatic speech recognition (ASR) and natural language understanding (NLU) components. However, an audio signal representing the user speech may also contain interfering sound generated by a media player that is playing audio content such as music. Before performing ASR and NLU, a system attempts to identify the content being played by the media player, such as by querying the media player or by analyzing the audio signal. The system then obtains the same content from an available source and subtracts the audio represented by the content from the audio signal.