Speech Extraction Workflow for Business Negotiation Reports

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing information processing systems struggle to efficiently summarize and analyze communication sessions, such as business negotiations, by accurately extracting relevant information from speech data to generate insightful reports.

Innovation Solution

Utilizing a large-scale language model (LLM) to convert speech into text, identify extraction items and groups, and generate business negotiation and analysis reports by aggregating extraction results from multiple sessions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If speech data from communication sessions is manually analyzed and summarized, then relevant information can be extracted, but the process is time-consuming and inefficient

Engineering Contradiction:
Improveinformation extraction efficiencyVSAvoidtime for manual analysis
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical analysis with automated speech recognition and natural language processing systems. The circuitry automatically transcribes speech to text data and extracts relevant information without human intervention, dramatically improving productivity while reducing time loss.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Loss of information

If all speech data from communication sessions is processed and stored, then complete information is available, but the data volume becomes unmanageably large

Engineering Contradiction:
Improvecompleteness of communication dataVSAvoidvolume of text data
Core Design Contradiction:
Loss of informationVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential and relevant information from complete speech data using predefined extraction items. The circuitry identifies and extracts specific utterance information related to particular topics or keywords, retaining only necessary data while discarding redundant content, thus maintaining information quality without storing excessive data volume.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the complete speech data into meaningful units based on extraction items and groups. The circuitry divides the continuous speech stream into discrete extractable elements organized by categories, making the data manageable and easier to process while preserving the essential information structure.

Inventive Principle:
Principle #1Segmentation

3Ease of operation

If generic summaries are generated from communication sessions, then overall content is captured, but specific insights and actionable information are lost

Engineering Contradiction:
Improvesimplicity of report generationVSAvoidspecificity of extracted insights
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

The patent applies different extraction criteria and analysis depths to different segments of the communication data based on local requirements. The circuitry identifies specific regions or topics within the speech data that require detailed extraction and applies targeted processing to those areas, ensuring that important specific insights are preserved while maintaining overall simplicity.

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS20260011324A1Information processing system, information processing apparatus, and information processing method
Publication Date: 2026.01.08 RICOH CO LTD
  • US20260011324A1 patent drawing
  • US20260011324A1 patent drawing
  • US20260011324A1 patent drawing

AI summary

An information processing system includes circuitry that: acquires speech uttered by a first participant and a second participant, the first participant and the second participant participating in a communication session; acquires text data from the speech through speech recognition; receives a setting of an extraction item to be extracted from the text data; and acquires an extraction result obtained by extracting first utterance information related to the extraction item from the text data.