Translated text and subtitle streaming display method, device and equipment

Abstract

The invention discloses a translation text subtitle streaming display method, device and equipment. The main idea is based on starting from the subtitle end and designing the streaming display of the translated text given by the translation engine. Specifically, the translated target text sent by the translation engine is stored in the preset subtitle buffer, and then the total number of words in the current buffer is detected. From this data, the corresponding word output speed strategy is adaptively configured, the so-called word output speed Strategy, that is, it can continuously extract and display the target text word by word from the subtitle buffer according to this strategy, so as to achieve the effect of streaming subtitle output. The present invention avoids the influence of the translation engine, can adapt to various language translation scenarios, and automatically adjusts based on the number of buffered words, and displays them one by one in units of words, thus avoiding the situation of displaying a large amount of text in a short time when the subtitles are output in real time , thereby effectively improving the subtitle display effect and improving the user's reading experience.

Description

technical field [0001] The present invention relates to the technical field of real-time subtitle processing, in particular to a method, device and device for streaming display of translated text subtitles. Background technique [0002] At present, artificial intelligence technology is developing rapidly, and speech recognition (speech transcription, dictation) and machine translation technologies have been widely used in many fields. Especially in real-time processing scenarios, such as conferences, teaching, speeches, etc., efficient recognition and translation can be performed while the speaker is speaking, and subtitles can be displayed online on screens and other carriers. [0003] At present, the subtitle display technology for recognition and transcription is relatively mature and perfect. For example, the speech transcription engine can transcribe the speech into text in real time, and the subtitle display control can dynamically follow the progress of the speaker's ...

AI Technical Summary

Problems solved by technology

But the translation scene is different. The translation process needs to recognize the voice first, then use the recognition results to translate, and then display the translated target text, that is, at least two stages of processing by the recognition engine and the translation engine , because there will be a certain delay between the target subtitles generated by this process and the speaking process of the speaker, so the existing real-time translation scenarios mostly output the translation results of the entire sentence of the translation engine directly, which leads to the current translation text ( That is, the target text) the subtitle display effect is not good, and the reading experience is poor

[0005] In this regard, although the above-mentioned defects can be solved theoretically by improving the processing capabilities of the engine (mainly the translation engine), it is difficult to cover such a huge demand because of the large number of languages ​​and dialects involved, especially if machine translation It is unrealistic to achieve the same efficiency as the recognition engine in terms of technical capabilities and resource costs

Images