Sensor-Based Target Sound Generation Without Wearable Microphones
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing devices struggle to acquire high-quality target sounds, particularly in environments where microphones cannot be mounted on sound sources like players during sports or performances, due to restrictions on device weight and unwanted voice leakage.
Innovation Solution
A learning device and method that utilizes sensor signals from multiple sensors, including microphones, acceleration, gyro, and positioning sensors, to generate high-quality target sound signals by performing machine learning on sensor data collected during training and rehearsal phases, allowing for the generation of object sound source signals without a physical microphone during actual performances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a microphone is mounted on a sound source (e.g., player) to acquire target sound, then the target sound quality is improved, but the device weight increases and causes physical load on the performer
Solution Approach 1:
The patent extracts the microphone from the wearable device configuration and places it in a stationary recording device. The recording device captures sounds in the environment, and through sound source separation technology, extracts the target sound (e.g., player's voice) from the recorded mixture of sounds. This eliminates the need for mounting microphones on players, reducing device weight while maintaining target sound acquisition capability.
2Measurement precision
If a microphone is mounted on a sound source to acquire target sound, then the target sound quality is improved, but unwanted voice leakage occurs
Solution Approach 1:
The patent removes the microphone from the player-wearable configuration, eliminating the source of voice leakage. Instead, a stationary recording device captures environmental sounds, and sound source separation technology extracts the target voice from the recorded signal, preventing unwanted voice leakage while maintaining target sound quality.
Solution Approach 2:
The patent introduces sound source separation technology as an intermediary process between sound recording and target sound extraction. The recording device captures a mixture of sounds, and the sound source separation algorithm acts as an intermediary to isolate the target voice from the mixture, enabling high-quality target sound acquisition without direct microphone contact with the player.
3Adaptability or versatility
If multiple sensors are mounted on an object to acquire sensor signals, then the signal processing capability is improved, but the device complexity increases
Solution Approach 1:
The patent extracts the complex multi-sensor processing functionality from the wearable device and relocates it to a separate learning device and recording device system. The recording device simply collects sensor signals, while the learning device performs the complex machine learning and sound source separation processing, reducing the complexity of the player-wearable device while maintaining advanced signal processing capability.
Data Source
AI summary
The present technique relates to a signal processing device, a signal processing method, a learning device, a learning method, and a program enabling acquisition of a target sound having high quality.A learning device including a learning unit configured to perform learning on the basis of one or a plurality of sensor signals acquired by one or a plurality of sensors mounted on an object and a target signal relating to the object and corresponding to a predetermined sensor and generate coefficient data configuring a generator having the one or the plurality of sensor signals as its inputs and having the target signal as its output. The present technique can be applied to a learning device.