Do I have to upload my video to generate subtitles?

No. On Sats4AI, your browser extracts the audio from the video locally — the video file itself never leaves your device. Only small compressed audio pieces are sent for transcription, then deleted.

Is there a length limit for AI subtitle generation?

No practical limit. Because the audio is processed in small pieces, a 2-hour movie works the same way as a 2-minute clip. You pay per minute of audio.

What subtitle formats do I get?

Both SRT (SubRip) and VTT (WebVTT) files, ready to use in VLC, media players, video editors, or the web. Timestamps come from the speech recognition model, cue by cue.

Which languages are supported?

13 spoken languages, including English, French, Spanish, German, Portuguese, Arabic, Hindi, Chinese, Japanese, Korean, Italian, Dutch, and Russian — with automatic detection.

How much does it cost to subtitle a movie with AI?

10 sats per minute of audio — a 90-minute movie is about 900 sats (usually under one US dollar, depending on the Bitcoin price). No account or subscription; if a job fails, the unused part is refunded automatically.

What happens if my file has several audio tracks?

You pick which one to subtitle. Files with multiple audio tracks (original language, dubs, commentary) show a track selector before payment, labeled with each track's language and format.

AI Subtitles for Any Video — Without Uploading It

The problem with every subtitle generator

Search for “AI subtitle generator” and every tool asks the same two things: create an account, and upload your entire video. For a 2-hour movie that means pushing gigabytes to someone else’s server, waiting, trusting them to delete it, and usually hitting a length or file-size ceiling anyway — or a subscription paywall sized for studios, not for one film.

That upload was never technically necessary. Subtitles only need the audio — about 1% of the file.

How this one is different

At sats4ai.com/subtitles, your browser does the heavy lifting locally. It opens the video on your machine, extracts the audio track, cuts it into small pieces at natural pauses, and sends only those compressed audio pieces for transcription. The pieces come back as timestamped text and are stitched into one subtitle file that covers the whole video.

Your video never leaves your device. Only audio is processed, piece by piece, and transcripts are deleted from our servers within 24 hours.
Any length. A 2-hour film works like a 2-minute clip — there is no upload ceiling because there is no upload.
Any common format. MP4, MKV, WebM, MOV — and audio-only files too (MP3, M4A...). Files with several audio tracks (original + dub + commentary) show a track picker before you pay.
Both formats. You download .srt and .vtt, ready for VLC, players, editors, or the web.
13 languages with auto-detection.
Interruption-proof. If your tab closes mid-job, re-select the same file — the job resumes where it stopped, without paying twice. Abandoned jobs are refunded automatically for the unused part.

What it costs

10 sats per minute of audio. A 90-minute movie is about 900 sats — usually under a dollar. One payment up front, from any Lightning wallet, no account. That’s the whole pricing page.

Why we built it this way

Sats4AI’s rule for every tool is the same: no signup, no identity, pay for exactly what you use. Subtitles turned out to be the perfect fit for the browser-first approach — the private thing (your video) stays with you, and the paid thing (speech recognition) is metered by the minute. The result is a subtitle generator with no length limit, no subscription, and no copy of your movie on anyone’s server, including ours.

For agents & developers

The browser flow above is for a person with a video open on screen. An AI agent or a batch script usually already has the audio, so it can skip the browser entirely and get the same timestamped .srt / .vtt straight from our L402 API and MCP server — paying per minute with Lightning, no account and no API key.

Caption a whole library. Point an agent at a folder of recordings and have it subtitle each one in a loop — no seats, no quota, just per-minute sats.
Captions inside a pipeline. An agent transcribing a meeting or podcast can ask for the SRT/VTT in the same paid step, then pass it to the next tool — translate the captions, summarize them, or publish them.

Agents call the transcribe_audio tool, which returns the subtitle files alongside the transcript. The full pay-per-call flow (send audio → 402 invoice → resend with the payment proof) is in the transcription guide, and every tool is listed on the L402 & MCP docs page.

Try it

Go to sats4ai.com/subtitles, pick a video, check the quoted price, pay with any Lightning wallet, and watch the transcript stream in. Works best in Chrome or Edge; recent Firefox and Safari also supported.