Rhyming Chat Bot for Speech Data Collection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional chat bot systems face challenges in gathering domain-specific and speaker-specific training data efficiently, leading to errors in speech recognition and requiring lengthy user interactions for data collection.

Innovation Solution

Implementing a rhyming game-based system that engages users interactively to collect speech data quickly, using a rhyming chat bot that analyzes voice inputs and provides feedback, thereby enhancing speech recognition accuracy and user engagement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If conventional chat bot systems collect speech data through traditional methods, then the data collection process is straightforward, but it requires lengthy user interactions and results in insufficient training data

Engineering Contradiction:
Improveamount of speech dataVSAvoidtime for data collection
Core Design Contradiction:
Quantity of substanceVSLoss of time

Solution Approach 1:

The patent introduces a rhyming game as an intermediary mechanism between the chat bot and user. This game serves as a mediator that naturally elicits speech data from users without requiring direct data collection requests, thereby increasing data quantity while reducing collection time

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system allows users to voluntarily participate in the rhyming game and provide speech data through their own engagement with the game mechanics. Users self-generate training data by attempting to rhyme with the bot, eliminating the need for lengthy structured data collection sessions

Inventive Principle:
Principle #25Self-service

2Measurement precision

If conventional chat bot systems use standard speech recognition, then the system is simple to implement, but it produces errors in speech recognition

Engineering Contradiction:
Improvespeech recognition accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary processing of speech data by converting it to text and identifying rhyme schemes before main analysis. This preliminary structuring of data improves subsequent speech recognition accuracy by providing context and patterns that help disambiguate recognized speech

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The chat bot provides feedback to users about their rhyming attempts, creating a loop that allows the system to learn from user corrections and improve speech recognition accuracy over time. The feedback mechanism helps refine the recognition model through iterative learning

Inventive Principle:
Principle #23Feedback

3Quantity of substance

If the chat bot engages users in lengthy interactions to collect data, then more training data can be gathered, but user engagement and experience deteriorate

Engineering Contradiction:
Improvespeech data collectedVSAvoiduser experience
Core Design Contradiction:
Quantity of substanceVSEase of operation

Solution Approach 1:

The system dynamically adjusts the difficulty and structure of rhyming challenges based on user performance and preferences. This dynamic adaptation keeps interactions engaging and enjoyable while naturally collecting speech data, preventing user fatigue and maintaining ease of operation

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes parameters such as rhyme complexity, topic constraints, and interaction patterns to maintain user engagement. By varying these parameters, the bot collects diverse speech data while keeping interactions interesting and easy for users

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10991382B2Voice synthesized participatory rhyming chat bot
Publication Date: 2021.04.27 MICROSOFT TECHNOLOGY LICENSING LLC
  • US10991382B2 patent drawing
  • US10991382B2 patent drawing
  • US10991382B2 patent drawing

AI summary

Among other things, embodiments of the present disclosure may be used to help train speech recognizers for improving generalized voice experience quality in a chat bot system. In some embodiments, the system provides users with games to play to increase user engagement with the chat bot system.