Out of Vocabulary Pattern Learning for Speech Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech recognition systems face challenges in adapting to recognize out-of-vocabulary words, which are not included in the predefined dictionary, leading to inefficiencies in handling unique terms or dialects.
Innovation Solution
A hybrid speech recognition method that decodes speech into graphemic transcriptions, identifies candidate out-of-vocabulary words based on sequence length and repetition, presents these to the user for validation, and trains the system to recognize valid words, thereby expanding the vocabulary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a predefined dictionary is used for speech recognition, then recognition speed is improved, but the system cannot recognize out-of-vocabulary words
Solution Approach 1:
The patent segments speech into sub-word units (phonemes or character n-grams) that can be independently decoded and recombined. This allows the system to handle out-of-vocabulary words by breaking them into recognizable sub-components, resolving the contradiction between using a fixed dictionary for speed and needing to recognize new words.
Solution Approach 2:
The system dynamically adapts its vocabulary by identifying candidate out-of-vocabulary words from sub-word sequences and incorporating user-validated words into the recognition system. This dynamic update mechanism allows the system to expand its vocabulary over time while maintaining the efficiency of the base dictionary.
2Adaptability or versatility
If the vocabulary is expanded to include more words, then adaptability is improved, but system complexity increases
Solution Approach 1:
The system performs preliminary decoding of speech into sub-word units and identifies candidate out-of-vocabulary words before final recognition. This preliminary processing allows the system to prepare potential new vocabulary items without immediately incorporating them, reducing the complexity burden of full vocabulary expansion.
Solution Approach 2:
The system uses user feedback to validate candidate out-of-vocabulary words before incorporating them into the vocabulary. This feedback mechanism ensures that only truly useful words are added, preventing unnecessary complexity from being introduced while maintaining adaptability.
3Measurement precision
If user feedback is collected for validation, then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The system applies user feedback validation selectively only to candidate out-of-vocabulary words rather than all recognized speech. This partial application of the validation process maintains accuracy for new words while avoiding the time penalty for routine vocabulary recognition.
Data Source
AI summary
A method for adapting a speech recognition system for out-of-vocabulary, comprising, decoding by a hybrid speech recognition a speech including out-of-vocabulary terms, thereby generating graphemic transcriptions of the speech with a mixture of recognized in-vocabulary words and unrecognized sub-words, while keeping a track of the decoded segments of the speech, determining in the transcription sequences of sub-words as candidate out-of-vocabulary words based on a first condition with respect to lengths of the sequences of sub-words and a second condition with respect to the number of repetitions of the sequences, audibly presenting to a user the candidate out-of-vocabulary words from the corresponding segments of the speech according to the track, and receiving from the user indications of valid words corresponding to audible presentations of the sequences of sub-words in the candidate out-of-vocabulary words, and training a speech recognition to additionally recognize the candidate out-of-vocabulary words, thereby adapting the speech recognition to recognize out-of-vocabulary words, wherein the method is performed on an at least one computerized apparatus configured to perform the method, and an apparatus for performing the same.