Phonetic Search Indexing for Voice Query Transcription Errors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice command systems struggle with accurately transcribing certain terms, leading to errors in natural language understanding and inadequate search results for media items, particularly when using phonetic representations.
Innovation Solution
A system that generates multiple phonetic representations of media item metadata and updates a phonetic search index based on use frequency, leveraging phoneme and heuristic analysis to improve query accuracy by prioritizing relevant phonetic representations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If an automatic speech recognition module is used to convert voice commands to text, then voice input can be processed, but transcription errors occur for certain terms that propagate to the natural language understanding module
Solution Approach 1:
The patent introduces a phonetic search index as an intermediary between the ASR module and NLU module. This index stores multiple phonetic representations (pronounced as word, letter-by-letter concatenation) of query terms and uses phonetic matching to bridge transcription errors. When a query is received, the system performs phonetic matching against the index to find the intended term even when ASR transcription is incorrect.
Solution Approach 2:
The patent changes the parameter of term representation from single text form to multiple phonetic representations. Each term in the phonetic search index is stored with multiple phonetic encodings (e.g., IPA representations for both word-level and letter-level pronunciation). This parameter change allows the system to match queries based on phonetic similarity rather than exact text matching, thereby resolving transcription errors.
2Measurement precision
If all possible phonetic representations are stored in the phonetic search index, then query accuracy improves, but storage space and processing power requirements increase significantly
Solution Approach 1:
The patent performs preliminary action by pre-generating and storing only the two most common phonetic representations for each term in the phonetic search index: (1) the phonetic representation when the term is pronounced as a whole word, and (2) the phonetic representation when letters are concatenated and pronounced individually. This preliminary selection based on usage frequency avoids storing all possible phonetic variations, thereby reducing storage space and processing requirements while maintaining high query accuracy.
3Reliability
If a phonetic search index is used to improve query matching, then transcription error correction is enabled, but system complexity increases
Solution Approach 1:
The patent segments the query processing into distinct stages: (1) ASR converts voice to text, (2) the query is converted to phonetic representation, (3) phonetic matching is performed against the phonetic search index, and (4) the matched term is passed to NLU. This segmentation isolates the phonetic matching functionality as a separate, manageable component, reducing overall system complexity while improving reliability.
Data Source
AI summary
Systems and methods are described for modifying a phonetic search index based on a use frequency associated with phonetic representations of text terms included in metadata of a media item. A first phonetic representation of a text term of the metadata, pronounced as a word, may be generated. A second phonetic representation of the text term may be generated by concatenating a phonetic representation of each letter in the text term. A database may be queried to determine use frequencies of the first and second phonetic representations, one of which may be selected based on a comparison of the use frequencies. A phonetic search index may be modified by including an entry for the selected phonetic representation. A voice query related to the media item may be received, and a reply to the voice query may be generated for output by performing a lookup in the modified phonetic search index.