Speech Segmentation Using Acknowledgement Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for transcribing telephone conversation contents fail to accurately extract and analyze speech information due to the lack of effective utilization of customer acknowledgments, leading to difficulties in dividing speech into appropriate utterances during the transcription process.
Innovation Solution
An apparatus and method that utilize acknowledgement responses to divide conversational dialog into utterances by detecting pause and acknowledgement locations, extracting boundary candidates, and segmenting speech using a word database and grammar database to improve recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional transcription methods are used without utilizing acknowledgements, then the transcription process is simpler, but speech cannot be accurately divided into utterances and recognition accuracy deteriorates
Solution Approach 1:
The patent segments the transcription process into distinct functional modules: acknowledgement detecting section identifies acknowledgement positions, pause detecting section finds pause locations, and boundary-candidate extracting section generates potential utterance boundaries. This segmentation allows each module to specialize in one task, improving overall accuracy while making the complex process more manageable and systematic
Solution Approach 2:
The patent introduces intermediary components that bridge different parts of the system: the acknowledgement detecting section acts as an intermediary to identify listener responses, the pause detecting section mediates between raw audio and boundary identification, and the boundary-candidate extracting section serves as an intermediary between detection results and final segmentation. These intermediaries enable accurate speech division without requiring complete redesign of the entire transcription system
2Measurement precision
If speech is not divided into appropriate utterances, then the processing is faster, but accurate recognition and analysis of speech cannot be achieved
Solution Approach 1:
The patent performs preliminary actions by detecting acknowledgements and pauses before final boundary determination. The acknowledgement detecting section and pause detecting section prepare boundary candidate information in advance, allowing the recognizing unit to efficiently process segmentation without performing exhaustive analysis during the main recognition phase. This preliminary preparation significantly improves recognition accuracy while minimizing time loss
Solution Approach 2:
The patent replaces traditional mechanical speech segmentation methods (which rely on simple threshold-based pause detection) with an acoustic model-based approach. The recognizing unit uses acoustic models and statistical methods to evaluate boundary candidates, substituting crude mechanical division with intelligent acoustic analysis that achieves high accuracy without excessive processing time
3Loss of information
If acknowledgements are not utilized in segmentation, then the method is simpler, but information on conversation contents cannot be accurately extracted
Solution Approach 1:
The patent implements feedback by having the acknowledgement detecting section continuously monitor the listener's channel and feed back acknowledgement positions to the boundary-candidate extracting section. This feedback loop ensures that acknowledgement information is systematically incorporated into the segmentation process, preventing information loss about conversation structure and participant turns while maintaining a clear and organized method
Data Source
AI summary
An apparatus, method and program for dividing a conversational dialog into utterance. The apparatus includes: a computer processor; a word database for storing spellings and pronunciations of words; a grammar database for storing syntactic rules on words; a pause detecting section which detects a pause location in a channel making a main speech among conversational dialogs inputted in at least two channels; an acknowledgement detecting section which detects an acknowledgement location in a channel not making the main speech; a boundary-candidate extracting section which extracts boundary candidates in the main speech, by extracting pauses existing within a predetermined range before and after a base point that is the acknowledgement location; and a recognizing unit which outputs a word string of the main speech segmented by one of the extracted boundary candidates after dividing the segmented speech into optimal utterance in reference to the word database and grammar database.