Acoustic Quality Evaluation Data Correction for Hands-Free Systems

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In loudspeaker hands-free communication systems, the evaluation of acoustic quality by comparing evaluation target and reference sounds is compromised by interference sounds, such as wraparound of the near-end talker's voice and ambient noise, which are superimposed on the reference sound, leading to inaccurate assessments, especially when the acoustic echo canceller is used.

Innovation Solution

A data correction apparatus is employed to identify and remove interference sounds from the reference sound by determining voiced sections that do not include the voice of the call partner and updating these sections with a non-voice signal, thereby correcting the test data used in acoustic quality evaluations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a listening test is used to evaluate acoustic quality in loudspeaker hands-free communication systems, then the evaluation can be performed objectively with higher reproducibility and shorter conduction time, but interference sounds such as wraparound of near-end talker's voice and ambient noise are superimposed on the reference sound, leading to inaccurate evaluation results

Engineering Contradiction:
Improveevaluation efficiencyVSAvoidevaluation accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by detecting and marking interference sound sections in the reference sound before performing the quality evaluation. The correction target determination unit identifies sections containing wraparound or ambient noise, and these sections are corrected beforehand to ensure accurate evaluation results without compromising evaluation efficiency

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts and removes interference sounds from the reference sound by identifying sections where wraparound or ambient noise is present. The correction execution unit replaces these interference sections with appropriately synthesized signals, thereby separating the harmful interference from the useful evaluation data

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If the acoustic echo canceller is used to remove acoustic echoes, then the acoustic echo cancellation performance is improved, but the wraparound of near-end talker's voice is removed from the evaluation target sound while being present in the reference sound, creating a mismatch that lowers evaluation accuracy

Engineering Contradiction:
Improveacoustic echo cancellation performanceVSAvoidevaluation accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent applies local quality by making different parts of the reference sound have different characteristics. Specifically, sections identified as containing wraparound are corrected to match the processed evaluation target sound, while other sections remain unchanged. This localized correction ensures that the reference sound accurately reflects what should be evaluated in each specific section

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS11924368B2Data correction apparatus, data correction method, and program
Publication Date: 2024.03.05 NIPPON TELEGRAPH & TELEPHONE CORP
  • US11924368B2 patent drawing
  • US11924368B2 patent drawing
  • US11924368B2 patent drawing

AI summary

To improve accuracy of an evaluation in an acoustic quality evaluation test performed by comparing an evaluation target sound and a reference sound. A data correction apparatus 3 compares, in a call performed between a near-end terminal 1 and a far-end terminal 2, an evaluation target sound in which a voice output from the near-end terminal 1 is recorded and a reference sound in which a voice spoken by a call partner using the far-end terminal 2 to correct test data used in a listening test for evaluating acoustic quality of the call. A correction target determination unit 31 determines, as a correction target section, a voiced section that does not include the voice of the call partner detected from an acoustic signal representing the reference sound. A correction execution unit 32 updates the correction target section of the acoustic signal representing the reference sound with a non-voice signal predetermined.