20240046949. REAL-TIME AUDIO PROCESSING SYSTEM, REAL-TIME AUDIO PROCESSING PROGRAM, AND METHOD FOR TRAINING SPEECH ANALYSIS MODEL simplified abstract (REALTEK SEMICONDUCTOR CORPORATION)
REAL-TIME AUDIO PROCESSING SYSTEM, REAL-TIME AUDIO PROCESSING PROGRAM, AND METHOD FOR TRAINING SPEECH ANALYSIS MODEL
Organization Name
REALTEK SEMICONDUCTOR CORPORATION
Inventor(s)
REAL-TIME AUDIO PROCESSING SYSTEM, REAL-TIME AUDIO PROCESSING PROGRAM, AND METHOD FOR TRAINING SPEECH ANALYSIS MODEL - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240046949 titled 'REAL-TIME AUDIO PROCESSING SYSTEM, REAL-TIME AUDIO PROCESSING PROGRAM, AND METHOD FOR TRAINING SPEECH ANALYSIS MODEL
Simplified Explanation
Abstract: This patent application describes an audio real-time processing system, program product, and method for training a speech analysis model. The speech analysis model is trained to obtain mask information from an original audio, which is used to mask the original audio and generate a target audio. The system then analyzes the target audio and the original audio to obtain multiple analyzed audio sections and identifies repeated audio sections. The repeated audio sections are outputted by the system.
- The patent application describes an audio real-time processing system, program product, and method for training a speech analysis model.
- The speech analysis model is trained to obtain mask information from an original audio.
- The mask information is used to mask the original audio and generate a target audio.
- The system analyzes the target audio and the original audio to obtain multiple analyzed audio sections.
- The system identifies repeated audio sections from the analyzed audio sections.
- The repeated audio sections are outputted by the system.
Potential Applications:
- Speech recognition and transcription systems
- Audio editing and remixing software
- Noise cancellation and audio enhancement technologies
Problems Solved:
- Efficiently training a speech analysis model to obtain mask information from an original audio
- Identifying repeated audio sections in real-time for further processing or analysis
Benefits:
- Improved accuracy and efficiency in speech analysis and processing
- Enhanced audio editing capabilities
- Better noise cancellation and audio enhancement results
Original Abstract Submitted
an audio real-time processing system, an audio real-time processing program product and method for training speech analysis model are provided. the speech analysis model is firstly trained to obtain, from an original audio, mask information which is used to mask the original audio to get a target audio. the system obtains a plurality of analyzed audio according to the target audio and the original audio, obtains repeated audio section according to the plurality of the analyzed and output the repeated audio section.
