Scene-Aware Audio Processing for Speech Intelligibility Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to accurately measure and adjust speech intelligibility in audio content due to sensitivity issues with microphones and lack of consideration for producing intentions, leading to inadequate speech transmission in noisy environments.
Innovation Solution
An electronic apparatus and method that separates speech and non-speech signals to identify intelligibility, adjusting gain and processing based on scene information to optimize speech clarity for different audio types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If non-speech volume is measured by tracking minimum value of power for each frequency band, then measurement is simplified, but non-speech that abruptly increases cannot be properly measured and accuracy deteriorates
Solution Approach 1:
The patent implements dynamic tracking of non-speech volume by continuously monitoring power values over time and detecting abrupt changes. Instead of using a static minimum value approach, the system adapts to changing audio conditions by identifying when non-speech power levels suddenly increase, allowing accurate measurement of both steady-state and transient non-speech components across frequency bands.
2Ease of operation
If speech probability is measured by frequency bands to adjust output parameters, then frequency-specific control is achieved, but speech and non-speech of the same band increase together causing incomplete speech enhancement
Solution Approach 1:
The patent segments speech and non-speech signals within each frequency band using voice activity detection. By separating these signal types at the spectral level, the system can independently control gain adjustment for speech components while preserving or attenuating non-speech components in the same frequency band, thereby achieving precise speech enhancement without inadvertently amplifying non-speech content.
3Device complexity
If only speech intelligibility enhancement operation is performed for all input content, then processing is simplified, but producing intention of producer according to audio content type is not reflected
Solution Approach 1:
The patent implements dynamic processing strategies that adapt to different audio content types. The system analyzes the characteristics of input content (such as music, speech, or mixed content) and selectively applies speech intelligibility enhancement operations only when appropriate. This allows the system to preserve the producer's artistic intent for musical content while providing speech enhancement for speech-dominant content, achieving versatility without excessive complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided are an electronic apparatus and a controlling method thereof. The electronic apparatus includes an inputter and a processor configured to, based on receiving an audio signal through the inputter, obtain a speech intelligibility for the audio signal, and modify the audio signal so that the speech intelligibility becomes a target intelligibility that is set based on scene information regarding a type of audio included in the audio signal, and the type of audio includes at least one of a sound effect, shouting, music, or a speech.