Gain Factor Attenuation for Wideband Speech Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing wideband speech coding technologies face challenges in efficiently extending narrowband speech coders to support wideband frequency ranges without significant bandwidth increase or computational complexity, particularly in mobile and embedded applications, and often require transcoding, which is impractical.

Innovation Solution

A method that calculates and attenuates gain factor values based on the time-varying relation between low-frequency and high-frequency envelopes of a speech signal, allowing for efficient wideband speech coding with minimal bandwidth increase and avoiding transcoding, by using a system with envelope calculators and gain factor attenuators to process narrowband and highband signals separately.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If wideband speech coding is implemented using narrowband coding techniques scaled to cover wideband spectrum, then wideband frequency range support is achieved, but computational complexity increases significantly and bandwidth increases unacceptably

Engineering Contradiction:
Improvewideband frequency range supportVSAvoidcomputational complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The speech signal is divided into narrowband portion (0-4 kHz) and highband portion (4-8 kHz). The narrowband portion is encoded using conventional narrowband coding techniques, while the highband portion is generated by extrapolating the spectral envelope from the narrowband spectral envelope. This segmentation allows wideband speech coding without requiring computationally intensive wideband CELP coding throughout the entire frequency range.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

A spectral envelope extrapolation mechanism serves as an intermediary to generate the highband spectral envelope from the narrowband spectral envelope. This intermediary approach allows the highband signal to be derived without direct measurement or intensive processing of the highband signal itself, reducing computational complexity while maintaining speech quality.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If wideband speech coding is implemented using narrowband coding techniques scaled to cover wideband spectrum, then wideband frequency range support is achieved, but bandwidth increases unacceptably

Engineering Contradiction:
Improvewideband frequency range supportVSAvoidbandwidth
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The highband signal is extracted and generated through spectral envelope extrapolation rather than being directly transmitted. By deriving the highband portion from the narrowband spectral envelope through mathematical extrapolation, the system avoids transmitting redundant highband information, thus minimizing bandwidth increase to only about 800-1000 bps.

Inventive Principle:
Principle #2Taking out (Extraction)

3Quantity of substance

If spectral envelope extrapolation is used to extend highband from narrowband spectral envelope, then bandwidth increase is minimized and transcoding is avoided, but spectral envelope accuracy deteriorates

Engineering Contradiction:
ImprovebandwidthVSAvoidspectral envelope accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The gain factor is made dynamic and time-varying rather than fixed. The gain factor values are calculated based on the time-varying relation between the narrowband signal envelope and highband signal envelope, allowing the system to adapt to changing speech conditions. This dynamic adjustment compensates for inaccuracies in spectral envelope extrapolation by optimizing the scaling of the extrapolated highband signal in real-time.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP1875464B9Method, storage medium and apparatus for gain factor attenuation
Publication Date: 2020.10.28 QUALCOMM INC
  • EP1875464B9 patent drawingFigure 1a~1b
  • EP1875464B9 patent drawingFigure 2a~2b
  • EP1875464B9 patent drawingFigure 3a~3b

AI summary

A method of signal processing according to one embodiment includes calculating an envelope of a first signal that is based on a low-frequency portion of a speech signal, calculating an envelope of a second signal that is based on a high-frequency portion of the speech signal, and calculating a plurality of gain factor values according to a time-varying relation between the envelopes of the first and second signal. The method includes attenuating, based on a variation over time of a relation between the envelopes of the first and second signals, at least one of the plurality of gain factor values. In one example, the variation over time of a relation between the envelopes is indicated by at least one distance among the plurality of gain factor values.