Virtual Speaker Residual Encoding for Lower HOA Audio Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for encoding and decoding multi-channel audio data, such as higher order ambisonics (HOA), result in large data amounts and high bandwidth occupation due to the need for direct encoding of each sound channel, which is inefficient and difficult to transmit and store.
Innovation Solution
An audio encoding method that selects a first target virtual speaker from a preset virtual speaker set based on a scene audio signal, generates a virtual speaker signal and a residual signal, and encodes these signals into a bitstream, reducing the amount of data by using attribute information to represent the sound field.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If direct encoding of each sound channel is used in HOA technology, then audio quality is maintained, but data amount increases and bandwidth occupation rises
Solution Approach 1:
The patent segments the HOA encoding process into two distinct parts: a core encoder that processes all sound channels to capture the main sound field information, and additional encoders that process only the difference information for specific virtual speakers. This segmentation allows the system to maintain audio quality while reducing the overall data amount by avoiding redundant encoding of identical information across all channels.
Solution Approach 2:
The patent extracts only the essential sound field information using a core encoder with a limited number of channels, and then takes out only the necessary difference information for specific virtual speakers. This extraction approach eliminates redundant data while preserving the critical audio quality elements, directly addressing the contradiction between maintaining audio quality and reducing data amount.
2Reliability
If the number of sound channels increases to capture more detailed sound scene information, then audio quality improves, but encoding and decoding complexity increases
Solution Approach 1:
The patent divides the encoding system into a core encoder that handles all channels uniformly and additional encoders that handle only specific virtual speakers. This segmentation reduces encoding complexity by avoiding the need for every encoder to process every channel, while still achieving high audio quality through the combination of core and additional encoded information.
Solution Approach 2:
The patent applies different encoding strategies to different parts of the sound field: a universal core encoding for all channels and speaker-specific additional encoding only where needed. This local quality approach optimizes the balance between audio quality and encoding complexity by concentrating computational resources only where they are most beneficial.
3Measurement precision
If all sound channels are encoded to ensure complete sound field representation, then sound field accuracy is maintained, but transmission and storage efficiency decreases
Solution Approach 1:
The patent extracts only the essential sound field information through core encoding and then extracts only the necessary difference information for specific virtual speakers. This extraction methodology maintains sound field accuracy by preserving critical information while dramatically improving transmission and storage efficiency by eliminating redundant data.
Solution Approach 2:
The patent merges the core encoded information with additional encoded information from selected virtual speakers to reconstruct the complete sound field. This merging approach ensures sound field accuracy is maintained while improving efficiency, as the system transmits and stores only the essential core information plus selective difference information rather than all channel data.
Data Source
AI summary
Audio encoding and decoding methods and apparatuses are disclosed, to reduce an amount of encoded and decoded data, so as to improve encoding and decoding efficiency. The method includes: selecting a first target virtual speaker from a preset virtual speaker set based on a first scene audio signal; generating a first virtual speaker signal based on the first scene audio signal and attribute information of the first target virtual speaker; obtaining a second scene audio signal using the attribute information of the first target virtual speaker and the first virtual speaker signal; generating a residual signal based on the first scene audio signal and the second scene audio signal; and encoding the first virtual speaker signal and the residual signal, to produce encoded signals, and writing the encoded signals into a bitstream.