Speech Coding Pitch Contour Codebook Bit Rate Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing speech encoding methods face challenges in efficiently encoding pitch lags, leading to inaccurate or unnatural pitch representation, which affects speech decoding, especially due to the high bit rate required for transmitting frequent pitch lag updates.

Innovation Solution

A method using a pitch contour codebook to encode pitch lags, where a pitch contour index and an average pitch lag are encoded with fewer bits, forcing the analysis to find pitch lags represented by vectors in the codebook, preventing unnatural pitch representations and reducing bit rate.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If pitch lags are updated frequently to improve accuracy, then pitch representation accuracy is improved, but bit rate increases

Engineering Contradiction:
Improvepitch lag accuracyVSAvoidbit rate
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The pitch lag information is segmented into two parts: average pitch lag (common to multiple subframes) and pitch contour deviations (specific to each subframe). This segmentation allows efficient encoding by separating the redundant average information from the unique contour information, reducing overall bit rate while maintaining accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The average pitch lag is calculated preliminarily across multiple subframes before encoding individual pitch lags. This preliminary calculation enables the use of relative encoding for deviations from the average, reducing the number of bits needed for each individual pitch lag while maintaining high accuracy.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If pitch lags are encoded with high precision to improve naturalness, then speech quality is improved, but bit rate increases

Engineering Contradiction:
Improvespeech naturalnessVSAvoidbit rate
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The encoding approach changes from absolute pitch lag values to relative deviations from an average pitch lag. This parameter transformation maintains the ability to represent natural pitch variations while significantly reducing the number of bits required, as deviations are typically smaller and can be encoded more efficiently.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The pitch contour codebook dynamically adapts to the average pitch lag of the current speech segment. By making the codebook content dependent on the average pitch lag, the system can efficiently represent natural pitch variations without requiring high bit rates, as the codebook is tailored to the specific characteristics of each speech segment.

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If pitch lag updates are performed at every subframe to improve accuracy, then pitch representation is improved, but complexity increases

Engineering Contradiction:
Improvepitch lag accuracyVSAvoidencoding complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

Multiple pitch lag measurements from consecutive subframes are merged into a single average pitch lag value, which is then used as a reference for encoding individual pitch lags. This merging reduces the number of independent encoding operations required, simplifying the encoding process while maintaining high accuracy through the use of relative deviations.

Inventive Principle:
Principle #5Merging (Combining)

4Measurement precision

If a large pitch lag codebook is used to represent all possible pitch lags, then pitch accuracy is improved, but memory requirements and processing complexity increase

Engineering Contradiction:
Improvepitch lag representation accuracyVSAvoidcodebook complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The pitch contour codebook is designed with local quality by tailoring its content to the specific average pitch lag of each speech segment. Instead of using a single large codebook for all possible pitch lags, the system uses smaller, adaptive codebooks that are optimized for the local characteristics of each segment, reducing memory requirements and processing complexity while maintaining high representation accuracy.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP2384506B1Speech coding method and apparatus
Publication Date: 2017.05.03 SKYPE
  • EP2384506B1 patent drawingFigure 1a~1b
  • EP2384506B1 patent drawingFigure 2a~2b
  • EP2384506B1 patent drawingFigure 3~4

AI summary

A method, program and apparatus for encoding speech. The method comprises: receiving a signal representative of speech to be encoded; at each of a plurality of intervals during the encoding, determining a pitch lag between portions of the signal having a degree of repetition; selecting for a set of said intervals a pitch lag vector from a pitch lag codebook of such vectors, each pitch lag vector comprising a set of offsets corresponding to the offset between the pitch lag determined for each said interval and an average pitch lag for said set of intervals, and transmitting an indication of the selected vector and said average over a transmission medium as part of the encoded signal representative of said speech.