Scalable Audio Encoder Bit Allocation for Speech Quality and Concealment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scalable coding methods face challenges in enhancing speech quality and data concealment without increasing transmission rates, particularly when frame loss occurs, as they often result in speech deterioration even without frame loss.
Innovation Solution
A scalable coding apparatus that includes a core layer coding section and an enhancement layer coding section, which generates quality-improving encoded data and concealment data, with adaptive bit allocation based on speech modes to prioritize either quality improvement or concealment, ensuring robustness without increasing overall bit rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If concealment information is added to enhancement layer encoded data, then error robustness is improved, but transmission rate increases
Solution Approach 1:
The patent dynamically adjusts the allocation of bits between quality improvement data and concealment data based on channel conditions and speech characteristics. The bit allocation is not fixed but adapts to varying requirements, allowing the system to optimize between quality enhancement and error robustness in real-time without permanently increasing transmission rate.
Solution Approach 2:
The patent changes the parameter of bit allocation distribution between different enhancement layer data types. By adjusting the proportion of bits allocated to concealment information versus quality improvement information, the system can achieve better error robustness when needed while maintaining efficient transmission rates under normal conditions.
2Reliability
If codes are reallocated from quality improvement to concealment, then error robustness is improved, but speech quality deteriorates when no frame loss occurs
Solution Approach 1:
The system dynamically switches between allocating bits to quality improvement and concealment based on whether frame loss is detected. When no frame loss occurs, bits are allocated to quality improvement for better speech quality. When frame loss is detected, the allocation shifts to concealment data for error robustness, thus avoiding permanent quality deterioration.
Solution Approach 2:
The patent implements feedback mechanisms where the decoder monitors for frame loss conditions and sends information back to the encoder about channel conditions. This feedback allows the system to adjust bit allocation in real-time, ensuring that speech quality is optimized when conditions are good while providing error robustness when losses occur.
3Ease of manufacture
If extrapolation-based concealment is used, then implementation is simple, but robustness to packet loss is insufficient
Solution Approach 1:
The patent prepares concealment information in advance during the encoding phase, creating redundant data that can be used for concealment before actual packet loss occurs. This preliminary preparation of concealment data allows the decoder to switch to using pre-prepared concealment information rather than attempting simple extrapolation when loss occurs.
Solution Approach 2:
The patent introduces an intermediary concealment information layer that mediates between the core layer and enhancement layer. This intermediary contains pre-encoded concealment data that can be activated when packet loss occurs, providing a more robust solution than direct extrapolation while maintaining simpler implementation than full redundancy encoding.
Data Source
AI summary
A scalable encoding device is capable of improving quality of a decoded signal without increasing an encoding amount and compensating data with a sufficient quality upon data loss. An extension layer bit distribution calculator calculates a bit distribution of a quality improving encoding data and compensation encoding data in the extension layer according to an audio mode of the input signal. An extension layer encoder generates quality improving encoding data according to the specified number of bits. A compensation information encoder extracts a part of core layer encoding data and makes it as compensation encoding data for the core layer. An extension layer encoded data generator multiplexes the extension layer bit distribution information, the compensation encoding data, and the quality improving encoding data so as to obtain extension layer encoding data.