Speaker-Specific Voice Amplification via Acoustic Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In noisy environments, such as public spaces or remote work settings, background noise interferes with voice conversations, making it difficult for individuals to communicate clearly without amplifying unwanted sounds.
Innovation Solution
A computing device generates a user-specific acoustic model based on a voice sample, allowing it to selectively amplify the user's voice during audio conversations while reducing background noise in real-time, using features like pitch variations, periodicity measurements, and spectral shape analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If background noise is amplified along with the user's voice to ensure clear communication, then the user's voice becomes more audible, but the background noise also becomes more prominent and interferes with the conversation
Solution Approach 1:
The audio signal is segmented into multiple frequency components using Fourier transform. The system processes different frequency bands separately, allowing selective amplification of voice frequencies while suppressing background noise frequencies. This segmentation enables precise control over which audio components are enhanced and which are reduced.
Solution Approach 2:
The acoustic model applies different processing characteristics to different frequency regions. Voice frequencies receive amplification with preserved quality, while background noise frequencies receive attenuation. The system adapts the filtering and amplification parameters locally for each frequency band based on the learned voice characteristics, achieving selective enhancement without uniform processing.
2Device complexity
If a generic acoustic model is used to amplify all voices equally, then implementation is simpler, but it cannot selectively amplify a specific user's voice in noisy environments
Solution Approach 1:
The system performs preliminary training to create a user-specific acoustic model before actual voice amplification. During this preliminary phase, the system learns the user's voice characteristics from training data, storing the learned parameters for later use. This preliminary action enables the system to quickly and accurately isolate the user's voice during real-time communication without complex processing during the actual conversation.
Solution Approach 2:
The system creates a simplified representation (acoustic model) that copies the essential characteristics of the user's voice. This copied model contains the learned voice patterns, pitch ranges, and spectral features, allowing the system to identify and amplify the user's voice without needing to process the entire complex audio signal in real-time.
3Object-affected harmful factors
If traditional noise cancellation methods are used, then background noise is reduced, but the user's voice may also be attenuated along with the noise
Solution Approach 1:
The system uses feedback from the acoustic model to guide the noise reduction process. The learned voice characteristics serve as a reference that continuously feedbacks to the filtering algorithm, allowing it to distinguish between voice and noise components. This feedback mechanism ensures that noise reduction operations preserve the user's voice while removing background interference.
Solution Approach 2:
The system replaces traditional mechanical or simple electronic noise cancellation methods with a data-driven acoustic model approach. Instead of using fixed filtering techniques that may attenuate voice signals, the system uses learned statistical models to intelligently distinguish and preserve voice components while removing noise, achieving better voice fidelity.
Data Source
AI summary
A method, system and computer program product for amplifying a single voice during an audio conversation. One embodiment of the method may comprise receiving, by a computing device, an audio sample of speech from a user, and generating, by the computing device, a user-specific acoustic model for enhancement of speech by the user based upon the audio sample. The method may further comprise receiving a live audiovisual stream, the live audiovisual stream including live speech by the user during an audio conversation, wherein the live audiovisual stream includes background noise, and using, by the computing device, the user-specific acoustic model to selectively amplify the live speech during the live audiovisual stream without amplifying the background noise.