Speech Analysis Using Prosodic Segments for Personality Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for determining behavioral and psychological characteristics from speech patterns are not situation-independent, as speech patterns are significantly affected by situational factors, making it difficult to measure personality traits consistently across varying contexts.

Innovation Solution

A computer-implemented method using secondary speech parameters such as rising-pitch, falling-pitch, and equal-pitch segments, which are less affected by individual differences, to analyze prosodic features and determine situational behavioral and psychological characteristics, enabling speaker-independent processing and identification of speech styles.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If speech patterns are used to measure personality characteristics, then behavioral and psychological characteristics can be determined, but the measurement is affected by situational factors and lacks consistency across different contexts

Engineering Contradiction:
Improvepersonality measurement consistencyVSAvoidsituational variability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent segments speech analysis into two distinct components: prosodic features (pitch, intensity, tempo, rhythm) and lexical/content features. By separating these elements, the system can focus on prosodic patterns that are more stable indicators of personality traits while minimizing the influence of situational factors that may affect speech content and style.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms raw speech signals into specific prosodic parameters (fundamental frequency, intensity, duration, spectral features) that serve as stable indicators of personality characteristics. These parameter transformations enable consistent measurement across different situations by focusing on acoustic properties that reflect underlying personality traits rather than situational responses.

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If prosodic features are analyzed to determine emotional state, then emotional information can be extracted, but individual differences and speech style variations interfere with accurate classification

Engineering Contradiction:
Improveemotional information extractionVSAvoidspeaker-independent processing
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent extracts and isolates specific prosodic features (pitch contours, intensity variations, temporal patterns) from the overall speech signal, separating the emotional information carriers from individual speech style characteristics. This extraction process enables focus on universal emotional indicators while filtering out speaker-specific variations.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent develops a universal prosodic analysis framework that can identify emotional states across different speakers and languages. By focusing on fundamental prosodic patterns that are common to human speech rather than speaker-specific characteristics, the system achieves emotional recognition that works across diverse individuals and contexts.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Reliability

If speech patterns are used to classify personality types, then behavioral characteristics can be identified, but the method requires sampling under very similar conditions to be reliable

Engineering Contradiction:
Improvepersonality classification reliabilityVSAvoidsampling condition constraints
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The patent employs dynamic baseline adjustment that adapts to each speaker's natural speech patterns and the specific context of analysis. Rather than requiring fixed sampling conditions, the system dynamically calibrates to the speaker's prosodic characteristics and situational context, enabling reliable personality assessment across varying conditions through adaptive normalization.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3396668B1Classifying speech utterances
Publication Date: 2021.07.28 VOICESENSE LTD
  • EP3396668B1 patent drawingFigure 1
  • EP3396668B1 patent drawingFigure 2
  • EP3396668B1 patent drawingFigure 3

AI summary

A computer implemented method of analyzing speech utterances of a speaker in a given situation and context and determining behavioral, psychological and speech style characteristics of the speaker in the given situation, said computer implemented method comprising: creating a speech parameters reference database for classifying speech utterances according to various behavioral, psychological and speech styles characteristics; obtaining speech utterances of a speaker in a specific situation and context; deriving a plurality of secondary speech parameters from said primary parameters; calculating a subset of speech parameters, parameters combinations and parameters' values representative of situational behavioral, psychological and speech styles characteristics, from said secondary parameters in the speech utterance; determining and scoring the situational behavioral, psychological and speech style characteristics in the speech utterance by comparing the calculated subset of speech parameters, parameters combinations and parameters' values with the pre-defined reference database of speech parameters.