Multilingual Speech Recognition Correction via Language Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition technologies are inadequate for accurately recognizing multilingual speech, as they often misinterpret words from different languages, leading to inaccurate transcription.
Innovation Solution
A method and apparatus that perform speech recognition by first obtaining a text through general-purpose ASR, correcting it using a mapping relationship between words in different languages, and outputting the corrected text if it corresponds to the same language, employing a machine translation model to ensure accurate multilingual speech recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If general-purpose ASR is used for speech recognition, then the system can process speech in a single language, but it fails to accurately recognize multilingual speech containing words from different languages
Solution Approach 1:
The system segments the speech recognition task into multiple language-specific recognition processes. It identifies different language segments within the input speech and routes them to appropriate language models for accurate recognition of multilingual content
Solution Approach 2:
The system creates a universal speech recognition framework that can handle multiple languages. By integrating multiple language models and using language identification techniques, the system achieves multi-functionality to process both monolingual and multilingual speech effectively
2Adaptability or versatility
If the system attempts to recognize all languages simultaneously, then it can handle multilingual speech, but the recognition accuracy decreases due to language confusion
Solution Approach 1:
The system dynamically adjusts the language model selection based on the detected language composition of the input speech. It uses language identification to determine which languages are present and activates only the relevant language models, avoiding the confusion that would arise from using all language models simultaneously
Solution Approach 2:
The system introduces language identification as an intermediary step between speech input and recognition processing. This mediator analyzes the speech to determine language composition and guides the selection of appropriate recognition models, ensuring accurate multilingual recognition without cross-language interference
Data Source
AI summary
A speech recognition method including performing speech recognition on an inputted speech to obtain a first text, correcting the first text according to an obtained mapping relationship between words in different languages to obtain at least one second text, and in response to determining that the at least one second text corresponds to the same language, outputting the first text, or in response to determining that the at least one second text corresponds to different languages, determine an outputted text according to first probability values corresponding to each of the at least one second text. By combining the mapping relationships between words in different languages in correcting the initial ASR result, the present application ensures the accuracy of the final speech recognition result.