Audio Decoder Bit Error Detection in Lattice Codebooks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio codecs, such as the Enhanced Voice Service (EVS) codec, face challenges in detecting bit errors in lattice vector quantized audio signals, particularly in circuit-switched networks where bit errors occur, due to the inability to determine invalid indexes with traditional power-of-two codebook cardinalities.
Innovation Solution
The method involves determining bit errors by analyzing index integer values, sub-indexes, and parameter component values within the lattice vector quantized data, and controlling the decoding process by setting codevectors to defined values, such as zero, to generate accurate audio signals despite bit errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional power-of-two codebook cardinalities are used in lattice vector quantization, then the decoding process is simpler and more efficient, but the ability to detect bit errors is lost
Solution Approach 1:
The patent applies asymmetry by using non-power-of-two codebook cardinalities (e.g., 15, 23, 31) instead of traditional power-of-two values. This asymmetric choice creates a detectable pattern: when bit errors occur, the calculated index values will exceed the valid range defined by the non-power-of-two cardinality, providing a reliable error detection mechanism. The asymmetric structure transforms the codebook size from a purely efficiency parameter into a dual-purpose feature that also enables error detection.
2Reliability
If bit error detection is implemented in lattice vector quantized audio signals, then audio quality is maintained in error-prone environments, but the decoding complexity increases
Solution Approach 1:
The patent implements preliminary action by performing bit error detection early in the decoding process, before the full audio reconstruction occurs. The method calculates index values from the received bitstream and checks them against the non-power-of-two cardinality constraints upfront. If errors are detected at this preliminary stage, the decoder can take corrective actions (such as requesting retransmission or using error concealment) before investing computational resources in the complete decoding process, thereby managing complexity effectively.
Solution Approach 2:
The patent introduces an intermediary error detection mechanism that sits between the received bitstream and the audio reconstruction process. This intermediary layer calculates index values and validates them against the non-power-of-two codebook constraints, acting as a mediator that prevents erroneous data from propagating through the complex audio decoding pipeline. This intermediary check adds minimal complexity compared to the full decoding process while significantly improving reliability.
3Reliability
If non-power-of-two codebook cardinalities are used, then bit error detection becomes possible through index validation, but the codebook design and implementation become more complex
Solution Approach 1:
The patent applies parameter changes by modifying the codebook cardinality parameter from traditional power-of-two values to non-power-of-two values (such as 15, 23, 31). This parameter change fundamentally alters the validation logic: instead of checking if indices are within a power-of-two range, the system checks against the specific non-power-of-two cardinality. This parameter modification enables error detection while the implementation complexity is managed through systematic algorithms for generating and validating these non-standard codebook structures.
Data Source
AI summary
A method comprising: receiving lattice vector quantised parameter data, the parameter data representing at least one audio signal; determining within the data at least one bit error; and controlling the decoding of the data to generate an audio signal based on the determining of the bit error.