Audio Processing System Using Re-trained Synthesis Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional audio signal processing techniques deteriorate sound quality when modifying sounding conditions such as pitches, volumes, and phonetic identifiers, leading to suboptimal editing results.
Innovation Solution
An audio processing method involving a re-trained synthesis model that generates feature data for acoustic features of an audio signal based on modified sounding conditions, using pre-trained models and additional training with specific condition and feature data from a first audio signal, allowing for precise modification of audio signals while maintaining sound quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional audio signal processing techniques are used to modify sounding conditions (pitch, amplitude, etc.), then the audio signal can be edited according to user instructions, but the sound quality deteriorates
Solution Approach 1:
The system performs preliminary analysis of the audio signal to extract acoustic features and sounding conditions before modification. By pre-processing the signal to identify pitch, amplitude, and other acoustic characteristics, the system prepares the data in a structured format that enables quality-preserving modifications later in the process
Solution Approach 2:
The patent introduces an intermediate representation layer between the original audio signal and the modified output. Acoustic features are extracted as intermediate data structures that serve as mediators, allowing modifications to be applied to the feature representation rather than directly to the raw signal, thereby preserving sound quality while enabling editing operations
2Measurement precision
If pitch and amplitude of audio signal are analyzed and displayed for editing, then user can modify audio parameters, but sound quality deteriorates due to modification of sounding conditions
Solution Approach 1:
The system implements feedback by continuously monitoring the acoustic features during the modification process. The extracted pitch and amplitude information feeds back into the synthesis model, allowing the system to adjust modifications in real-time to maintain sound quality while achieving the desired editing results
Solution Approach 2:
Instead of directly modifying the raw audio signal, the system changes parameters in the acoustic feature domain. By adjusting pitch and amplitude as separate extracted features and then reconstructing the signal from these modified features, the system achieves precise parameter control without the quality degradation that occurs with direct signal manipulation
Data Source
AI summary
An audio processing system and a method thereof generate a synthesis model that can input an audio signal to generate feature data that can be used by a signal generator to generate a modified audio signal. Specifically, a pre-trained synthesis model is first generated using training audio data. Thereafter, a re-trained synthesis model is established by additionally training the pre-trained synthesis model. Based on a received instruction to modify at least one of sounding conditions of an audio signal to be processed, feature data is generated by inputting additional condition data into the re-trained synthesis model. The signal generator generates the modified audio signal from the generated feature data.