安德烈·卡帕西谈智能体工程
一视频，通世界
Bringing foreign content in or taking your own content out — it's the same pipeline. Preview English→Chinese free without signing up, no card — hear the first 10 minutes for yourself. For Chinese→English, register and create with free trial credits.
DROP A VIDEO, OR CHOOSE A FILE
MP4 · MOV · M4V · WebM · up to 30 min
No signup · No card · Sign in only when you want the file
Rather hear it first? Jump to real samples
Both ways
More language pairs rolling out
10min
Free preview length
180min
Longest single video
CapCut
Editable draft project
One-click tools collapse six jobs into one button, so when something breaks you have nowhere to start. We keep them apart: every intermediate result is readable, editable and re-runnable, and a fix only recomputes what sits downstream of it.
Isolate speech, separate speakers, produce a timecoded source script.
Accuracy first, with a glossary to lock terms and proper nouns.
Rework each line to the target language's spoken rhythm so it can actually be said.
Cloned or preset voice, generated line by line.
Two-way micro-adjustment against timecode, so drift never accumulates.
Export the finished video, or hand off a CapCut draft for further cutting.
Isolate speech, separate speakers, produce a timecoded source script.
SRT exportAccuracy first, with a glossary to lock terms and proper nouns.
GlossaryRework each line to the target language's spoken rhythm so it can actually be said.
Pace budgetCloned or preset voice, generated line by line.
Voice libraryTwo-way micro-adjustment against timecode, so drift never accumulates.
Drift visibleExport the finished video, or hand off a CapCut draft for further cutting.
CapCut draftEvery clip below is a real render, untouched afterwards. Listen for the end-of-sentence pauses and where the emotion lands — that's where one-click tools give themselves away. All current samples are English→Chinese finished cuts; for Chinese→English, sign up and hear it yourself with your trial allowance.
安德烈·卡帕西谈智能体工程
YouTube · 96jN2OCOfLs · AI Ascent 2025 · 16:25 – 17:18 · 52s
Translation, pace, voice and timecode all line up on one row. Editing a line re-synthesises only that line; everything else stays exactly as it was. Export the finished video, or hand off a CapCut draft for further cutting.
Line-by-line review
Source, translation and dub side by side; risky lines flagged
Incremental re-render
Only affected segments re-run — credits aren't spent twice
CapCut draft handoff
An openable project, not a flattened MP4
Live product UI · Studio line review
Both promise "upload a video, get another language." The difference shows up the moment you're not happy with it.
Preview the first 10 minutes without an account. Subscribe when you like what you hear — the checkout price is authoritative, and failed jobs are never charged.
Sign up to get a 7-day trial with trial credits (about 20 minutes of processing) and a job slot plus Studio fine-tuning mode. No automatic charge when the trial ends, and your account info and purchased credits are kept indefinitely.
Run one job end to end
Volume and long-form
A lot of knowledge content isn't a few-second short — it's a 30-minute, 1-hour, or even 3-hour interview, course, podcast, or talk. Our Pro plan supports a single video up to 180 minutes long, built exactly for translating and dubbing this kind of long-form content in either direction.
Both directions are supported: turn English long videos into Chinese-dubbed versions, or your Chinese videos into English-dubbed versions — interviews, courses, podcasts, and product demos work most reliably. The no-sign-up preview currently supports English → Chinese only; for Chinese → English, sign up and pick the direction when creating a task. Other languages (Japanese, Spanish, etc.) are under evaluation and not yet available — please check with support before uploading.
In Studio mode, Chinese-to-English jobs support line-by-line translation review and voice selection during generation; Express runs fully automatically. The post-completion Edit entry (line-level re-dubbing and finished-cut edits) is currently English-to-Chinese only. For Chinese-to-English, we recommend finishing your review in the Studio flow before exporting.
We support importing your own or duly authorized content: local video file upload, YouTube links (for your own channel or videos you have rights to), and other video links. Before using, please confirm you hold the legal rights to the video — see the Terms of Service for details.
Rask, HeyGen, and ElevenLabs are all excellent global platforms, some very strong on digital avatars or voice cloning. AITrans.Video's focus is different — long-form translation and dubbing between Chinese and English, in both directions: longer supported video durations, a lower barrier to entry, multiple export formats, and the ability to review sentence by sentence and regenerate single segments (the post-completion edit entry currently supports English-to-Chinese only).
What the AI generates is a first draft, not the final version. In the workbench you can review the translation, subtitles, and dubbing sentence by sentence, edit any sentence you're not happy with, and regenerate just that segment.
No. AITrans.Video supports incremental regeneration: when you change just one sentence, the system reprocesses only that segment instead of re-synthesizing the whole video, saving both time and cost.
Express mode runs the whole pipeline automatically — fast and low-cost, good for validating results first or producing in batches. Studio mode lets you review the translation sentence by sentence in the workbench and choose a more fitting dubbing voice, with more stable quality, better suited for content you'll publish or deliver to clients. With proper legal authorization, Studio also supports voice cloning.
Once a job is done you can download: the dubbed video, dubbed audio, subtitles (original / translated / bilingual), translated text, the raw materials pack, and a Jianying draft project — open the finished subtitle, dubbing, and material tracks directly in Jianying to keep fine-editing, without laying out the timeline from scratch.