Transfer Function Estimation for Unknown Microphone Arrays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech-processing technologies require prior knowledge of the microphone array arrangement and sound source positions to estimate transfer functions, making it difficult to obtain transfer functions from unknown microphone arrays and sound sources.
Innovation Solution
A speech-processing apparatus and method that uses a microphone array with unknown arrangements and unknown sound sources to estimate transfer functions by detecting speech zones, calculating feature quantities, and clustering to determine the number of sound sources, allowing for transfer function estimation without pre-emitted signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a predetermined signal is output from a microphone to dynamically estimate a transfer function, then the transfer function can be estimated, but a known speech signal must be output from a loudspeaker which is not always available
Solution Approach 1:
The system uses the actual speech signal from the speaker themselves to estimate the transfer function, rather than requiring an external predetermined signal. The speech processing apparatus processes the speaker's own speech to obtain transfer function information, making the system self-sufficient and eliminating the need for external signal generation equipment.
2Measurement precision
If geometric calculation or measurement of a specific signal is used to obtain transfer function, then transfer function information can be obtained, but it requires prior knowledge of microphone array arrangement and sound source positions
Solution Approach 1:
The patent replaces geometric calculation methods with signal processing methods. Instead of using physical measurements and geometric relationships between microphones and sound sources, the system uses speech signal processing and statistical methods to estimate transfer functions, eliminating the need for precise physical arrangement knowledge.
Solution Approach 2:
The system changes the approach from spatial parameter-based methods (requiring microphone positions and sound source locations) to signal-based methods. By using speech signal characteristics and statistical analysis, the system obtains transfer function information without relying on physical arrangement parameters.
3Adaptability or versatility
If a microphone array with unknown arrangement and unknown number of sound sources is used, then flexibility and ease of deployment are improved, but transfer function estimation becomes difficult
Solution Approach 1:
The speech processing apparatus autonomously estimates the number of sound sources and obtains transfer function information by processing the speech signals themselves, without requiring external input about the array configuration or sound source positions. The system serves itself by extracting all necessary information from the speech signals.
Solution Approach 2:
The patent introduces speech signal processing as an intermediary method between the microphone array and transfer function estimation. By using statistical and signal processing techniques on the speech signals, the system bridges the gap between unknown array configuration and the need for transfer function information.
Data Source
AI summary
A speech-processing apparatus includes: a representative transfer function estimation unit that uses a sound signal which is collected by using a microphone array of which the arrangement is unknown, which has a plurality of channels, and of which the number of sound sources is unknown and that estimates a transfer function with respect to a sound source.