Automatic Translation Apparatus Using Back-Translation Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic translation and interpretation systems face challenges in reproducing translated text in a user's own voice and accurately translating text with typing or orthographic errors, especially in noisy environments or with proper nouns of low frequency, leading to increased recognition and translation errors.
Innovation Solution
An automatic translation and interpretation apparatus that includes a speech input unit, text input unit, sentence recognition unit, translation unit, speech output unit, and text output unit, which extracts speech features, measures similarity between words, and converts uttered sounds into text in the original language, allowing for accurate translation and interpretation, even with errors, and outputs the translated text in the user's voice.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automatic translation is performed based on speech recognition results, then translation functionality is provided, but recognition errors increase translation errors
Solution Approach 1:
The system performs back-translation by translating the translated text back to the source language and comparing it with the original speech recognition result. This feedback mechanism identifies recognition errors by detecting discrepancies between the original and back-translated text, allowing the system to correct errors and improve translation accuracy while maintaining full automation.
Solution Approach 2:
The system performs speech-to-text conversion and back-translation before final translation output. This preliminary action of converting speech to text, translating back, and comparing allows error detection and correction to occur before the final translation is generated, improving overall translation accuracy while maintaining automation.
2Extent of automation
If speech recognition is performed in noisy environments or with low-frequency proper nouns, then speech-to-text conversion is provided, but recognition performance deteriorates
Solution Approach 1:
The system uses back-translation as a feedback mechanism to verify speech recognition accuracy. By translating the recognized text back to the source language and comparing with the original speech input, the system can identify and correct recognition errors that occur in noisy environments or with low-frequency proper nouns, thereby improving reliability while maintaining automated speech recognition.
3Ease of operation
If text input is accepted with typing errors, then text input flexibility is provided, but translation accuracy decreases
Solution Approach 1:
The system applies back-translation to text input by translating the input text back to the source language and comparing it with the original input. This feedback mechanism identifies typing or orthographic errors, allowing the system to correct them before final translation, thereby maintaining translation accuracy while preserving text input flexibility.
4Speed
If only forward translation is performed, then translation speed is maintained, but error correction capability is lost
Solution Approach 1:
The system performs back-translation as a preliminary action before final translation output. This additional step enables error detection and correction while maintaining relatively fast translation speed through automated processing. The back-translation comparison quickly identifies errors that can be corrected algorithmically, preserving speed while improving accuracy.
Data Source
AI summary
The present invention relates to an automatic translation and interpretation apparatus and method. The apparatus includes a speech input unit for receiving a speech signal in a first language. A text input unit receives text in the first language. A sentence recognition unit recognizes a sentence in the first language desired to be translated by extracting speech features from the speech signal received from the speech input unit or measuring a similarity of each word of the text received from the text input unit. A translation unit translates the recognized sentence in the first language into a sentence in a second language. A speech output unit outputs uttered sound of the translated sentence in the second language in speech. A text output unit converts the uttered sound of the translated sentence in the second language into text transcribed in the first language and outputs the text.