Switchable Audio Coder for Speech and Generic Signal Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech and audio coding technologies face challenges in processing generic audio inputs, such as music and reverberant speech, due to difficulties in differentiating between speech and generic audio signals, leading to inconsistent sound quality and distortion when using hybrid core codecs.
Innovation Solution
A method for classifying audio frames into speech-like and generic audio frames, processing each type with appropriate codecs, and using an enhancement layer to improve coding efficiency by transforming error signals into the frequency domain, with LPC coefficients used for perceptual weighting when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a classifier is used to select between speech and generic audio coders, then speech quality is improved, but frequent switching between coders occurs when signals are near the switching threshold, resulting in inconsistent sound quality
Solution Approach 1:
The patent implements dynamic coder selection by processing signals through multiple coder paths simultaneously and selecting the best output in real-time. The system dynamically adjusts between speech and generic audio coders based on signal characteristics, avoiding frequent switching by using a hybrid approach that combines both coding methods.
Solution Approach 2:
The patent creates a composite coding system that combines speech coding and generic audio coding techniques into a unified hybrid codec. This composite approach allows the system to leverage the strengths of both coding methods, processing speech-like signals with speech coders and music-like signals with generic audio coders, while maintaining consistent quality through perceptual weighting.
2Productivity
If an audio transform domain enhancement layer is used on top of a speech coder output, then coding efficiency is improved, but when generic audio signals are used as input, the output becomes distorted and substantial portion of enhancement layer coding effort goes to reversing noise produced by signal model mismatch
Solution Approach 1:
The patent applies preliminary classification to determine whether the input signal is speech-like or generic audio before selecting the appropriate coding path. By performing this classification beforehand, the system avoids applying speech coding enhancement layers to generic audio signals, thereby preventing the signal model mismatch and associated distortion that would otherwise require substantial enhancement layer effort to correct.
3Quantity of substance
If speech coders are used for generic audio inputs, then bit rate is reduced, but the output signal becomes distorted due to signal model mismatch
Solution Approach 1:
The patent dynamically changes coding parameters based on signal type classification. When generic audio signals are detected, the system switches from speech coding parameters to generic audio coding parameters, including different transform domains and coding schemes. This parameter adaptation allows efficient compression of generic audio while avoiding the signal model mismatch that causes distortion in speech-coded generic audio.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for processing an audio signal including classifying an input frame as either a speech frame or a generic audio frame, producing an encoded bitstream and a corresponding processed frame based on the input frame, producing an enhancement layer encoded bitstream based on a difference between the input frame and the processed frame, and multiplexing the enhancement layer encoded bitstream, a codeword, and either a speech encoded bitstream or a generic audio encoded bitstream into a combined bitstream based on whether the codeword indicates that the input frame is classified as a speech frame or as a generic audio frame, wherein the encoded bitstream is either a speech encoded bitstream or a generic audio encoded bitstream.