The smart subtitle feature can perform Automatic Speech Recognition (ASR) on the voice information in on-demand video files or live streams, convert it into subtitles, and translate it into multiple languages. It is suitable for scenarios such as live subtitles for streaming and video translation for overseas distribution. You can create custom smart subtitle templates and preset different process parameters for different use cases, to facilitate subsequent reuse.
Go to Templates > Media AI Template > Smart Subtitle. The system provides several preset templates that you can use directly, or you can click Create Smart Subtitle Template to create a custom template.
Enter the smart subtitle template creation page. The following configuration parameters are supported:
Configuration Item
Description
Template Name
Only Chinese characters, English letters, digits, underscores (_), hyphens (-), and periods (.) are supported, and the length should not exceed 64 characters.
Source Language of Video
Select the source language of the source video's audio. Currently, the following languages are supported:
Simplified Chinese, English, Japanese, Korean, Chinese-English-Cantonese, Medical Chinese, Cantonese, Vietnamese, Malay, Indonesian, Filipino, Thai, Portuguese, Turkish, Arabic, Spanish, Hindi, French, German, Italian, Russian, and Chinese dialects.
Among them, the following languages are supported for enabling the translation feature:
Commonly used words in speech are generally accurately recognized. However, for specific names (such as personal names, product names, and company names), and industry-specific terms (such as the brand name Zhilin, the building name Binhai Dasha, Hebao in the insurance field, or Cunchutong in cloud storage), the recognition accuracy may decrease. To address this issue, we provide a custom hotword lexicon feature. You can add specialized terms through manual input or file import, significantly improving the accuracy of ASR. For detailed configuration guidance, see the Custom Hotword Lexicon documentation.
Note:
Currently, the hotword lexicon only supports Mandarin and English languages. Therefore, the hotword lexicon can only be associated when the source language of the video's audio is Simplified Chinese or English.
Enable Translation
After this feature is enabled, the subtitles in the source language will be translated into the specified language, making it suitable for scenarios such as video translation for overseas distribution.
If the translation feature is not enabled, the output subtitles will only contain content in the source language.
When the translation feature is enabled, the following subtitle output types are supported:
Subtitles in the target language.
Subtitles in both the source language and the target language.
Subtitle File Format
Currently, the WebVTT format is supported. If you only need a subtitle content callback and do not require an actual subtitle file to be generated, you can select "Do Not Generate Subtitle Files".
