Media
5 items
Description
On-device multilingual emoji suggestions for YouTube subtitles powered by Emo model. In Short: Instant emojis react to YouTube video content. Powered by 100% on-device AI model 'Emo'. Detailed Description: Real-time floating emojis for YouTube videos. Powered by the offline Emo AI model. AI Model 'Emo' (by Desert Ant Labs): - Runs an 11MB LiteRT model + tokenizer locally in your browser worker. - Sub-2ms inference speed per subtitle line. - 22 languages supported natively (including CJK, English, Japanese, Korean). - 812-emoji vocabulary tuned specifically for short, intent-carrying text. - Converts line-by-line subtitle dialogue into Bilibili-style floating emojis in real time. No cloud backend: - No network requests initiated: Zero outbound fetch or XHR calls during AI inference. - No network latency: Instant local execution without server roundtrips. - No costs or API keys: Completely free, offline, and self-contained. - No data tracking: Subtitle text never leaves your browser sandbox. How to use: Open any YouTube video with captions. Emojis float automatically with dialogue. Desert Ant Labs: github.com/Desert-Ant-Labs