Voice Filtering Using Speaker Embeddings for Audio Clarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In noisy environments, users face difficulties in transmitting clear audio communications due to background noise, which existing voice filtering technologies struggle to effectively remove, especially in multi-user settings and real-time communications.
Innovation Solution
A method and system for voice filtering that utilize a voice filtering recognition routine to determine when to activate voice filtering based on contextual inputs, generating enhanced audio data by isolating the user's voice characteristics using a speaker embedding, and transmitting this enhanced data to ensure clear audio communication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If voice filtering is applied to remove background noise, then audio clarity is improved, but device complexity increases
Solution Approach 1:
The patent introduces a voice filtering engine as an intermediary component that sits between the audio input and the communication output. This engine receives raw audio data, applies voice filtering processing using speaker embeddings to isolate the user's voice, and outputs filtered audio data. By positioning this specialized processing layer as an intermediary, the system achieves effective background noise removal while maintaining a clear separation of concerns and modular architecture, thus managing device complexity systematically.
Solution Approach 2:
The voice filtering system is segmented into distinct functional components: a voice filtering recognition routine that determines when filtering should be applied, a voice filtering engine that performs the actual filtering, and speaker embedding modules that provide voice characteristics. This segmentation allows each component to be optimized independently and enables selective activation of filtering based on environmental conditions, improving audio clarity without requiring the entire system to operate at maximum complexity at all times.
2Measurement precision
If voice filtering is activated continuously, then audio communication quality is improved, but energy consumption increases
Solution Approach 1:
The system dynamically adjusts the activation state of voice filtering based on real-time conditions. The voice filtering recognition routine continuously monitors environmental factors and determines whether filtering should be active. This dynamic approach allows the system to activate filtering only when background noise levels warrant it, rather than running continuously, thus maintaining high audio communication quality when needed while reducing energy consumption during quiet periods.
Solution Approach 2:
The system changes the operational parameter of voice filtering activation based on environmental assessment. When the recognition routine detects high background noise levels, it changes the parameter to activate filtering; when noise levels are low, it changes the parameter to deactivate filtering. This parameter-based control enables the system to adapt to varying acoustic conditions, optimizing the balance between audio quality and energy usage.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A method (400) includes receiving a first instance of raw audio data (102) corresponding to a voice command (118) and receiving a second instance of the raw audio data corresponding to an utterance of audible contents (126) for an audio-based communication (150) spoken by a user (102). When a voice filtering recognition routine (200) determines to activate voice filtering for the voice of the user, the method also includes obtaining a respective speaker embedding (318) of the user and processing, using the respective speaker embedding, the second instance of the raw audio data to generate enhanced audio data (152) for the audio-based communication that isolates the utterance of the audible contents spoken by the user and excludes at least a portion of the one or more additional sounds that are not spoken by the user. The method also includes transmitting the enhanced audio data to a recipient device (105).