Audio Processing With Virtual Voice Parts for Single-Take Harmony
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of obtaining audio without accompaniment chorusing is high as users need to sing multiple times to achieve harmonies of different voice parts, increasing the complexity of audio obtaining.
Innovation Solution
An electronic device generates a second vocal of one or more voice parts associated with a first vocal, allowing users to play a target audio that includes the first and second vocals without the need for multiple harmony singing sessions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If users sing multiple times to obtain harmonies of different voice parts, then the quality of audio without accompaniment chorusing is improved, but the complexity of audio obtaining increases
Solution Approach 1:
The patent uses voice conversion technology to create virtual copies of the user's voice in different voice parts (e.g., tenor, alto, soprano). Instead of requiring the user to sing multiple times in different voice parts, the system converts the original vocal recording into multiple voice part variations, significantly reducing the complexity of obtaining audio without accompaniment chorusing while maintaining quality
Solution Approach 2:
The system performs preliminary voice conversion processing to pre-generate multiple voice part versions from a single original vocal recording. This preliminary action eliminates the need for users to perform multiple singing sessions, as all voice part variations are already prepared and can be combined directly to create the final audio without accompaniment chorusing
2Reliability
If users sing multiple times to obtain harmonies of different voice parts, then the completeness of voice parts is improved, but the time required for audio obtaining increases
Solution Approach 1:
The voice conversion system generates complete sets of different voice parts (tenor, alto, soprano, etc.) from a single original vocal recording by creating virtual voice copies. This ensures all necessary voice parts are obtained while dramatically reducing the time required, as the conversion process is automated and does not require multiple separate singing sessions
Solution Approach 2:
The system maintains continuous useful action by processing the original vocal recording once to generate all required voice part variations simultaneously. This continuous processing ensures completeness of all voice parts while minimizing time loss, as the entire voice conversion process occurs in a single operational flow rather than requiring multiple discrete singing sessions
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A method, apparatus and electronic device for audio processing are provided, and the method includes: obtaining a first vocal (S201); generating, based on the first vocal, a second vocal of one or more voice parts associated with the first vocal, a timbre of the second vocal being a predetermined timbre (5202); and playing a target audio based on the first vocal and the second vocal of the one or more voice parts, the target audio comprising the first vocal and/or the second vocal of the one or more voice parts (S203).