Automatic transcription with word-level timing
Every job starts with speech recognition that returns a timestamp for each word. Those timings drive when each subtitle chunk appears and which word is highlighted, so subtitles stay in sync without manual timing work.
Burned-in subtitles instead of sidecar files
Instead of returning an SRT or VTT file you then have to attach or render yourself, the API renders the subtitles into the video frames. The output MP4 plays with subtitles everywhere, including platforms that ignore or strip subtitle tracks.
Languages
The spoken language is detected automatically from the audio, so the same request works for English, Spanish, Portuguese, Hindi, Arabic and the other supported languages without any extra parameter.
Traditional subtitles or animated captions
Pick a sentence-level style for classic subtitles, or a short-chunk animated style for TikTok, Reels and Shorts. Every style accepts the same customization options for font size, colours, stroke, background and position.
