Pronunciation Alias Retrieval for Under-Locatable Media Names
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice and natural language understanding systems struggle to accurately recognize and locate media content items with names containing special characters, unconventional spellings or pronunciations, or those that sound similar to conventional words, leading to difficulties in accessing the desired content.
Innovation Solution
A system and method for detecting under-locatable media content items by analyzing playback history and user interactions, generating pronunciation aliases through crowd-sourcing, and classifying name entities to improve locatability using a retrieval engine and alias generation engine.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional speech-to-text processing is used to retrieve media content, then the system can process standard words and phrases, but it fails to accurately recognize names with special characters, unconventional spellings, or pronunciations that sound like other words
Solution Approach 1:
The system pre-generates pronunciation aliases for media content names before retrieval operations. When a user searches, the system checks multiple pre-computed pronunciations simultaneously, allowing accurate matching of unconventional names without requiring real-time complex processing.
Solution Approach 2:
The patent introduces pronunciation aliases as an intermediary layer between the user's speech input and the media content database. These aliases act as mediators that bridge the gap between conventional speech-to-text processing and unconventional name recognition, enabling accurate retrieval without modifying the core speech recognition system.
2Reliability
If the system uses exact transcription matching, then it can accurately retrieve content with conventional spellings, but it fails to retrieve content when users pronounce words differently due to accents, dialects, or unconventional spellings
Solution Approach 1:
The system transforms media content names into multiple phonetic parameter representations (pronunciation aliases) stored in the database. During retrieval, the user's speech is converted to text and matched against these varied parameter representations, allowing the same content to be retrieved regardless of pronunciation variations from accents or dialects.
3Adaptability or versatility
If the system expands its vocabulary to include all possible pronunciations and spellings, then it can recognize more unconventional names, but the system complexity and processing requirements increase significantly
Solution Approach 1:
Instead of generating all possible pronunciations for every media content item, the system selectively generates pronunciation aliases based on detected patterns from user search behavior. When the system observes users searching for content using alternative pronunciations, it automatically generates and adds relevant pronunciation aliases, implementing partial action only where needed.
Solution Approach 2:
The system automatically monitors user search patterns and self-generates pronunciation aliases based on observed misrecognitions. When users repeatedly search for content using pronunciations that don't match the stored name, the system autonomously creates new pronunciation aliases and updates its database without requiring manual intervention or system reconfiguration.
Data Source
AI summary
Methods, systems, and related products that provide detection of media content items that are under-locatable by machine voice-driven retrieval of uttered requests for retrieval of the media items. For a given media item, a resolvability value and/or an utterance resolve frequency is calculated by a number of playbacks of the media item by a speech retrieval modality to a total number of playbacks of the media item regardless of retrieval modality. In some examples, the methods, systems and related products also provide for improvement in the locatability of an under-locatable media item by collecting and/or generating one or more pronunciation aliases for the under-locatable item.