Cloud Conference Voice Control Using Frequency-Domain Speaker Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice call systems face delays in speaker quantity detection, especially in real-time communication scenarios, leading to reduced efficiency and accuracy in voice recognition and control.
Innovation Solution
A method involving frequency domain transformation and neural network-based separation processing to determine the energy proportion of each branch voice, allowing real-time detection and control of the number of speakers in a voice call.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a speaker scenario detection system is trained based on a large quantity of labeled voice segments using traditional methods, then the system can predict the quantity of speakers, but it causes a relatively long delay for voice detection which greatly reduces the efficiency of voice recognition and affects the effect of real-time voice control
Solution Approach 1:
The patent segments the voice signal into multiple frequency bands and processes each band separately to determine energy proportions. This segmentation allows parallel processing of different frequency components, reducing overall detection time while maintaining accuracy in speaker quantity detection.
Solution Approach 2:
The patent performs frequency domain transformation and energy proportion calculation in advance during the voice call, before final speaker quantity determination is needed. This preliminary processing prepares the data structure for rapid speaker quantity detection without requiring heavy computational resources at the moment of detection.
2Measurement precision
If traditional speaker detection methods are used, then speaker quantity can be predicted, but the processing manner causes a relatively long delay which greatly reduces the efficiency of voice recognition
Solution Approach 1:
By dividing the voice signal into frequency bands and processing them separately, the system can efficiently determine speaker quantity without requiring full-signal analysis, thus improving voice recognition efficiency while maintaining detection accuracy.
Solution Approach 2:
The patent uses energy proportion information from frequency bands as a partial feature for speaker quantity detection, rather than analyzing all voice characteristics. This partial action approach reduces computational complexity and improves processing efficiency while still achieving accurate detection.
3Productivity
If real-time voice control is implemented, then voice call efficiency is improved, but detection delay greatly affects the effect of real-time voice control
Solution Approach 1:
The system performs frequency domain transformation and energy proportion determination as preliminary actions during the voice call setup and ongoing processing. This prepares the detection framework in advance, enabling rapid real-time voice control responses without significant detection delay.
Solution Approach 2:
The patent maintains continuous frequency domain analysis and energy proportion monitoring throughout the voice call. This continuous useful action ensures that speaker quantity information is always available for real-time voice control without requiring periodic re-detection, eliminating detection delay.
Data Source
AI summary
Embodiments of this application provide a real-time voice call control method performed by an electronic device. The method includes: obtaining a mixed call voice in real time during a cloud conference call, where the mixed call voice includes at least one branch voice; determining energy information corresponding to each frequency point of the call voice in a frequency domain; determining an energy proportion of each branch voice at each frequency point in total energy of the frequency point based on the energy information at the frequency point; determining a quantity of branch voices comprised in the call voice based on the energy proportion of each branch voice at each frequency point; and controlling the voice call by setting a call voice control manner based on the quantity of branch voices.