Two-Stage Speech/Music Classification for Low-Bitrate Stereo Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio codecs face challenges in efficiently transmitting stereo sound signals at low bit-rates while maintaining sound quality, as they often double the bit-rate and fail to exploit redundancy between left and right channels, and do not support immersive audio experiences effectively.
Innovation Solution
A two-stage speech/music classification method and device that classifies input sound signals and selects a core encoder for encoding, using a Gaussian Mixture Model (GMM) for initial classification and additional high-level features for refined encoder selection, optimizing encoder switching for seamless transitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If both left and right channels of stereo sound signal are coded using a mono codec, then stereo sound transmission is achieved, but the bit-rate is doubled and redundancy between channels is not exploited
Solution Approach 1:
The patent merges the encoding of left and right channels by coding their sum (primary channel) and difference (secondary channel) together using a single mono codec, rather than encoding them separately. This combining approach exploits the redundancy between channels and reduces the total bit-rate required for stereo transmission.
Solution Approach 2:
The patent segments the stereo signal into two components: the sum of left and right channels (primary) and the difference between them (secondary). This segmentation allows the encoder to focus bits on the primary channel which contains most of the information, while using fewer bits for the secondary channel that represents the stereo difference.
2Quantity of substance
If a very low bit-rate is used for each of the left and right channels, then the overall bit-rate is kept at a reasonable level, but the sound quality is affected
Solution Approach 1:
The patent applies local quality by allocating different bit-rates to different channels based on their importance. The primary channel (sum of L and R) receives a higher bit-rate to maintain good sound quality, while the secondary channel (difference between L and R) uses a lower bit-rate since it contains less critical information for perceived quality.
3Manufacturing precision
If stereo parameters are coded and transmitted as signaling information, then stereo image information is preserved, but the overall bit-rate increases
Solution Approach 1:
The patent extracts the essential stereo information by computing the sum and difference of left and right channels, then codes only these two derived signals. This extraction approach preserves the stereo image while avoiding the need to transmit separate full-resolution left and right channels plus additional stereo parameters, thereby reducing the overall bit-rate.
4Productivity
If speech/music classification is performed to select appropriate core encoder, then encoding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent performs preliminary speech/music classification before the actual encoding process to determine which type of core encoder to use. By performing this classification in advance, the system can select the most appropriate encoder for the given signal type, improving encoding efficiency while managing complexity through structured pre-processing.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Two-stage speech/music classification device and method classify an input sound signal and select a core encoder for encoding the sound signal. A first stage classifies the input sound signal into one of a number of final classes. A second stage extracts high-level features of the input sound signal and selects the core encoder for encoding the input sound signal in response to the extracted high-level features and the final class selected in the first stage.