STAC Audio Codec Integer-Reversible MLT Transcoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional audio compression techniques face challenges such as inefficiency in predicting periodic tones, high computational overhead due to side information requirements, and reduced compression performance when dealing with large audio block sizes, especially in transcoding between lossless and lossy formats.
Innovation Solution
The STAC Codec employs an integer modulated lapped transform (MLT) for transforming audio signals into the frequency domain, combined with a backward-adaptive run-length Golomb-Rice (RLGR) encoder for compressing transform coefficients, allowing for efficient lossless and near-lossless compression with reduced computational overhead and enabling fast transcoding between formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If forward-adaptive prediction is used for audio compression, then compression can be achieved, but periodic tones cannot be efficiently predicted and high-order predictors require excessive computation
Solution Approach 1:
The patent replaces the time-domain predictive coding mechanism with a frequency-domain transform coding mechanism. Instead of using complex time-domain predictors to predict periodic tones, the invention applies a transform (such as MDCT) to convert the audio signal into the frequency domain, where periodic tones manifest as distinct spectral peaks that can be efficiently encoded without requiring high-order prediction. This substitution of the coding domain fundamentally resolves the contradiction between compression efficiency and predictor complexity.
2Productivity
If transform coding is used for lossless compression, then transcoding can be performed in the frequency domain, but integer-reversible transforms accumulate rounding errors for large block sizes
Solution Approach 1:
The patent changes the parameter of the transform from conventional integer-reversible transforms to a float-point transform with controlled rounding. Specifically, the invention uses a transform such as MDCT that operates in floating-point arithmetic with carefully controlled rounding operations, allowing large block sizes (e.g., 2048 or 4096 samples) to be processed without the accumulation of significant rounding errors. This parameter change enables both fast transcoding and high reconstruction accuracy simultaneously.
Solution Approach 2:
The patent implements prior cushioning by using overlap-add or overlap-save methods with extended block sizes. The transform is applied to overlapping blocks with appropriate windowing functions, and the rounding errors are distributed and canceled across overlapping regions. This beforehand cushioning approach prevents the accumulation of rounding errors even when using large block sizes for efficient transcoding.
3Quantity of substance
If conventional lossy compression is used, then storage requirements are reduced, but full decoding is required for transcoding operations
Solution Approach 1:
The patent implements a universal coding framework where the same transform-based encoder can operate in both lossless and lossy modes. The encoder uses a transform (such as MDCT) followed by either lossless entropy coding or lossy quantization, depending on the desired output. This multi-functionality allows the system to store audio in a lossless format when high quality is needed, while also enabling fast transcoding to lossy formats by simply adjusting the quantization parameters without requiring full time-domain decoding. The intermediate transform coefficients serve as a universal representation for both modes.
Data Source
AI summary
A “STAC Codec” provides audio transcoding and decoding by processing an encoded audio signal using a backward-adaptive run-length Golomb-Rice (RLGR) decoder to recover transform coefficients of the encoded audio signal. The transform coefficients are then either transcoded in the transform domain to lossy or other formats, or decoded to the time domain by applying an inverse integer-reversible modulated lapped transform (MLT) to the recovered transform coefficients to recover an uncompressed time domain representation compressed audio signal. In additional embodiments, an inter-block spectral estimation and inverse data sorting strategy is used in recovering the transform coefficients from the encoded audio signal. In other embodiments, conversion from lossless encoding to near-lossless encoding is achieved by right-shifting recovered transform coefficients by some number of bits such that quantization errors are not perceived as distortion in the decoded audio signal, then re-encoding the right shifted transform coefficients.