Speech Recognition Vocabulary Augmentation via Preliminary Text Transcription
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Speech processing systems struggle to identify new words, acronyms, names, and field-specific terms not included in their vocabulary, requiring impractical reprocessing of audio streams to update their lexicon, which is computationally expensive and inefficient.
Innovation Solution
A computer-implemented method that accepts new vocabulary items and generates an augmented characterization of audio signals by identifying putative occurrences of the new terms, updating the vocabulary, and reprocessing the audio stream to include these terms, allowing for efficient retranscription or analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the speech processing system reprocesses the audio stream to add new terms to the vocabulary, then the vocabulary coverage is improved, but the computational cost and processing time increase significantly
Solution Approach 1:
The system performs preliminary speech recognition on the audio stream to generate a transcript before the user submits the new term. This preliminary processing creates a searchable text representation that enables rapid verification of the new term's presence without requiring full reprocessing of the audio stream when the new term is added to the vocabulary.
2Measurement precision
If the speech processing system reprocesses the audio stream to search for new terms, then the detection accuracy is improved, but the time consumption increases
Solution Approach 1:
The system creates a text copy (transcript) of the audio stream content through preliminary speech recognition. This text copy serves as a surrogate that can be rapidly searched for new terms using simple string matching algorithms, achieving accurate term detection without the time-consuming process of reprocessing the original audio stream.
3Speed
If the speech processing system uses a closed vocabulary set, then the processing speed is maintained, but the ability to identify new words and specialized terms is lost
Solution Approach 1:
The system performs preliminary speech recognition to generate a transcript before new terms are added to the vocabulary. This allows the system to maintain its closed vocabulary processing speed for standard terms while enabling rapid identification and integration of new terms through efficient text-based verification of the transcript.
Data Source
AI summary
Systems, methods, and apparatus, including computer program products for accepting a predetermined vocabulary-dependent characterization of a set of audio signals, the predetermined characterization including an identification of putative occurrences of each of a plurality of vocabulary items in the set of audio signals, the plurality of vocabulary items included in the vocabulary; accepting a new vocabulary item not included in the vocabulary; accepting putative occurrences of the new vocabulary item in the set of audio signals; and generating, by an analysis engine of a speech processing system, an augmented characterization of the set of audio signals based on the identified putative occurrences of the new vocabulary item.