Wideband Speech Coding Split-Band Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current speech coding technologies are limited in transmitting wideband voice communications efficiently, particularly in extending the frequency range beyond traditional PSTN limits, which affects intelligibility and quality, especially for applications like VoIP and cellular telephony.

Innovation Solution

A wideband speech coding system employing a split-band coding scheme that separates the frequency range into narrowband and highband components, using different coding modes and rates for each band to efficiently encode and decode speech signals, allowing for transmission of a wider frequency range without significant quality loss.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional speech coding is used to limit bandwidth to 300-3400 Hz, then transmission efficiency is maintained, but speech intelligibility and quality deteriorate due to loss of high-frequency information

Engineering Contradiction:
Improvespeech intelligibilityVSAvoidfrequency range
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The speech signal is divided into two separate frequency bands: narrowband (300-3400 Hz) and highband (3400-8000 Hz). Each band is processed independently with appropriate coding schemes, allowing efficient transmission of the complete wideband signal by segmenting the frequency spectrum into manageable portions that can be encoded separately and recombined at the receiver.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If wideband frequency range (50 Hz to 8 kHz) is transmitted, then speech quality and intelligibility are improved, but transmission bandwidth requirements increase

Engineering Contradiction:
Improvespeech qualityVSAvoidtransmission bandwidth
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The wideband frequency range is segmented into narrowband and highband portions, each transmitted using optimized coding schemes. The narrowband portion uses traditional efficient codecs while the highband portion uses specialized highband extension techniques, allowing the system to achieve wideband quality without transmitting the entire wideband range at full resolution simultaneously.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different quality levels and coding approaches are applied to different frequency regions. The narrowband portion receives standard coding treatment while the highband portion receives enhanced processing tailored to its specific characteristics, optimizing overall speech quality while managing bandwidth consumption efficiently through region-specific optimization.

Inventive Principle:
Principle #3Local quality

3Quantity of substance

If different coding modes are used for active and inactive frames, then average bit rate is reduced, but complexity of frame classification and coding selection increases

Engineering Contradiction:
Improveaverage bit rateVSAvoidcoding mode selection
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The speech coder dynamically switches between different coding modes based on the activity detection of each frame. For inactive frames, a simplified low-bit-rate coding mode is used, while for active frames, a higher-bit-rate mode is employed. This dynamic adaptation allows the system to optimize bit rate efficiency by matching the coding complexity to the actual speech content requirements of each frame.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP2047464B1Systems, methods, and apparatus for wideband encoding and decoding of active frames
Publication Date: 2013.11.06 QUALCOMM INC
  • EP2047464B1 patent drawingFigure 1
  • EP2047464B1 patent drawingFigure 2
  • EP2047464B1 patent drawingFigure 3A~3B

AI summary

Applications of dim-and-burst techniques to coding of wideband speech signals are described. Reconstruction of a highband portion of a frame of a wideband speech signal using information from a previous frame is also described.