AI Video Subtitle Generator
Made for creators: auto-generate subtitles for any video — burned into the frame or as SRT/VTT, translatable into 11 languages
Speech recognition runs on our own server (Qwen3-ASR model with forced-alignment timestamps) — audio is transcribed and deleted within 30 minutes. Transcription takes roughly 20-50% of the video length; the optional burned-in video adds a re-encode on top.
⚠️ Please don't close or refresh this page until it's done, or you may lose the result.
💡 Uploads are chunked and resumable - the pace is set by your own network. A failed chunk retries by itself; you are only ever charged when the job actually completes.
✏️ Subtitle editor
Fix any line or timestamp (00:00:01,000 format), then Apply — downloads, translation and burning all use your edited version.
Pipeline stages and styling options
- ASR pass — transcription with word-level timings; punctuation added automatically.
- Translation pass (optional) — line-aware translation respecting reading-speed limits per cue.
- Burn pass — font, size, outline and position render into the pixels.
- Soft-sub option — mux SRT as a selectable track instead of burning, keeping video untouched.
Transcribe, translate and burn in one job
Speech recognition produces timestamped SRT/VTT in the spoken language, optionally translates line-by-line, then burns styled subtitles into the video.
Real speech recognition
Your audio or video is transcribed by Qwen3-ASR on our server - not a flaky browser speech engine - producing subtitles with accurate timing, ready to edit the moment processing finishes.
SRT, VTT, ASS or plain text
Download the transcript in the format your editor, platform or player expects. Subtitles stay synced to the original audio, so you can polish wording, split lines and export again at any time.
Translate or burn in one step
Hand the finished SRT to DeepSeek for a translation into eleven languages, or burn subtitles directly onto the video. Each optional step is charged separately, and only when you use it.
How it works
- 1
Upload your video (any format, up to 1 GB) — pick the spoken language or let Qwen3-ASR auto-detect it.
- 2
The server transcribes the speech and, by default, burns the captions straight into the frame.
- 3
Download the subtitled video ready to post — plus SRT/VTT for your editor, with optional AI translation into 11 languages.
- 4
Pick burn-in or soft subtitles, adjust styling, and download — the SRT itself is always downloadable too.
Frequently asked questions
How accurate are the subtitles?
We run Qwen3-ASR with word-level forced alignment for the timestamps — clear speech typically transcribes with excellent accuracy across many languages including English and Chinese. Heavy accents, background music or crosstalk lower accuracy; the SRT preview makes a quick manual pass easy.
What formats come out, and will timestamps survive translation?
You get standard SRT (works in every editor and player) plus WebVTT for web players. AI translation rewrites only the text lines — every cue keeps its exact timing, so the translated file drops straight into YouTube or your editor.
Is my video kept or used for training?
No. The upload, the extracted audio and the transcript are all deleted from the server within 30 minutes, and the job is only reachable with your session's signed token. Nothing is shared or used for training.
Related tools
What Are AI Video Subtitles?
AI video subtitles are captions generated automatically from the speech in your video — no manual typing, no transcription service. This tool runs Qwen3-ASR, a Whisper-class speech recognition model, on our VPS: upload any video up to 1 GB, get accurately timed captions, and either download them burned into the video ready to post, or as SRT and WebVTT files for your editor. Captions can also be AI-translated into 11 languages. It is built for YouTubers, TikTokers, course creators and anyone who wants more reach and accessibility from their videos. You can also edit the captions in the page before exporting, so the final file matches your wording, not just the model choice.
What this tool can do
- 🎙️ Qwen3-ASR transcribes speech with word-level timing
- 🌍 Source language hints for 10 languages, plus auto-detection
- 🔥 Burn subtitles directly into the video, ready to publish
- 📄 Export as SRT or WebVTT for any editing software
- 🌐 AI-translate captions into 11 languages
- ✏️ Edit captions in the page, then re-burn the video
When you would use it
- You publish videos and want captions for reach and engagement
- You want your content accessible to deaf and hard-of-hearing viewers
- You repurpose a video into other languages for new audiences
- You need subtitle files for a course or an internal video library
Your video is uploaded to our VPS, where Qwen3-ASR transcribes it; the files are auto-deleted within 30 minutes. Transcription consumes credits based on duration, and each AI translation run costs a separate small fee. Honest about quality: recognition depends on the audio itself — heavy background noise, strong accents, overlapping speech or very fast talking all reduce accuracy, so we provide an in-page editor for review and correction before you export.
Related tools
Extract Audio from Video
Pull the soundtrack out of any video as lossless WAV — right in your browser
Video Frame Extractor
Grab still frames from any video as PNG or JPEG — right in your browser
Screen Recorder
Record your screen, window, or tab — with mic and system audio — free, no install, no upload