Perso Dubbing · Agent Skill

AI Subtitle Translator: Extract and Translate Subtitles from Any Video

AI Subtitle Translator: Extract and Translate Subtitles from Any Video

AI Subtitle Translator: Extract and Translate Subtitles from Any Video

Perso Dubbing’s subtitle translator turns videos into subtitles with speech recognition covering 100 languages, then hands the SRT to your AI agent to translate, so you pay speech-to-text rates only, with no dubbing fee. Point it at a file, a folder, or a URL, and translated subtitle files come back saved next to your originals.

Free trial · No credit card required · Works with any AI agent that runs skills

Folder-scale batch: web takes 1 file per job

No cap

Video length: web caps at 30 minutes

Any URL

URL sources: web takes YouTube, TikTok, Drive

0

Download clicks: files save next to originals

How it works

From video to translated subtitles, in one instruction

The /srt skill runs the whole pipeline. You describe the job in plain language; the agent handles every step.

01

Point at your media

video · folder · URL

Give the agent a file, a whole folder, or a link: YouTube, TikTok, Drive, or any direct URL.

02

Extract subtitles

100-language STT

Speech recognition covering 100 languages turns the audio into an accurate original-language SRT.

03

Your agent translates

agent model · 0 credits

Your AI agent translates the SRT into every language you asked for, inside its own model, with no Perso credits used.

04

Files land beside originals

.srt × languages

One SRT per language, saved next to your source videos and organized, with zero download clicks.

See it in action

Watch the agent do it

A short walkthrough of the skill extracting subtitles from a folder of videos and translating them, with no manual uploads and no downloads.

0:00

/

0:00

0:00

/

0:00

0:00

/

0:00

Agent vs web

Six ways the agent skill wins

Compared with the Perso Dubbing web workspace.

∞ vs 1

∞ vs 1

Whole folders at once

The web takes one file per job (2 GB). The skill mixes folders, files, and URLs, so every video × language pair runs through one batch.

No limit vs 30 min

No limit vs 30 min

Any video length

Long recordings are split losslessly, processed, and merged back. The web editor caps a single video at 30 minutes; the agent runs as long as your credits last.

Any URL vs 3

Any URL vs 3

Links from anywhere

The web reads links from YouTube, TikTok, and Drive; the agent also takes any direct video URL. File uploads work in mp4, mov, webm, mp3, wav, and m4a.

5 min

5 min

Unattended retries

When the queue is full, the skill re-checks every 5 minutes and resumes interrupted jobs from checkpoints. Leave it overnight, with no double billing.

0 clicks

0 clicks

Downloads handled

Results are saved next to your source files, organized by language. The web hands you per-file or ZIP downloads instead.

STT only

STT only

No dubbing fee for subtitles

The skill runs speech-to-text only; your agent translates the SRT itself. On the web, translated SRTs require a full dubbing project.

Deep dive

Translated subtitles, without the dubbing fee

Both paths end with translated SRT files. What you pay for along the way is very different.

Web workspace

A dubbing project, end to end

A dubbing project, end to end

1. Create a dubbing project. The full pipeline runs, voice generation included.

2. Wait for completion, then open the download menu.

3. Collect the translated SRT as a by-product of the dub.

Speech-to-text alone only produces the original-language transcript, so translated subtitles aren’t available without dubbing.

Cost: full dubbing rates

Agent · /srt skill

Straight to the subtitles

Straight to the subtitles

1. The skill creates a speech-to-text project only. No voices are generated.

2. The original-language SRT comes back.

3. Your agent translates it into every language you asked for, with no Perso credits consumed.

One SRT per language, saved next to the source video and organized for you.

Cost: speech-to-text rates only

Feature by feature

The bold side is the stronger one.

Feature

Feature

Web workspace

Web workspace

Agent skill

Agent skill

Input

Input

1 file per job · 2 GB

1 file per job · 2 GB

Folders, files, and URLs, mixed in one request

Folders, files, and URLs, mixed in one request

Video length

Video length

Up to 30 minutes per video

Up to 30 minutes per video

No length cap: limited only by your credits

No length cap: limited only by your credits

URL sources

URL sources

YouTube · TikTok · Drive

YouTube · TikTok · Drive

Plus any direct video link

Plus any direct video link

Queueing

Queueing

Manual, one job at a time

Manual, one job at a time

Global video × language queue, 5-minute auto-retry

Global video × language queue, 5-minute auto-retry

Interruptions

Interruptions

Upload retry only

Upload retry only

Checkpoint resume with a double-billing guard

Checkpoint resume with a double-billing guard

Results

Results

Per-file or ZIP download

Per-file or ZIP download

Auto-saved next to originals, organized by language

Auto-saved next to originals, organized by language

Translated subtitles

Translated subtitles

Via a full dubbing project, at dubbing rates

Via a full dubbing project, at dubbing rates

Speech-to-text rates: your agent translates free

Speech-to-text rates: your agent translates free

Segment editing

Segment editing

Edit lines, re-translate, assign speakers, preview

Edit lines, re-translate, assign speakers, preview

Not available

Not available

Voice control

Voice control

3 TTS models, voice cloning, up to 10 speakers

3 TTS models, voice cloning, up to 10 speakers

Default engine, single speaker

Default engine, single speaker

Where the web editor still wins

No overclaiming: for precision work, the workspace remains the better tool. Both run on the same account and the same credits.

Segment editing

Edit source or translated lines, re-translate a single sentence, assign speakers, and preview audio before export.

Voice control

Three TTS models, a voice-cloning library, and up to 10 speakers per video for dubbing work.

Extra inputs and outputs

Upload your own SRT, apply custom dictionaries, burn in subtitles, and export VTT, timestamps, or per-speaker sheets.

Resolution tiers

Choose 1080p or 4K output directly in the editor interface.

Rule of thumb: batch and unattended work belongs to the agent, and precision editing belongs to the web.

Rule of thumb: batch and unattended work belongs to the agent, and precision editing belongs to the web.

Up and running in three steps

No new interface to learn. Your agent does the clicking.

01

Install the skill and connect your key

Add the Perso Dubbing skill to your AI agent and connect your API key from developers.perso.ai.

02

Point it at your videos

A file, a folder, or a URL. Then ask in plain language: “extract subtitles and translate them into Spanish and Japanese.”

03

Collect your SRT files

The original transcription plus one SRT per target language, saved next to your source videos.

Subtitle translator FAQ

How do I translate subtitles from a video?

Install the Perso Dubbing skill in your AI agent, connect your API key, and point the agent at a video file, folder, or URL with the languages you want. It extracts subtitles with speech recognition covering 100 languages, translates them, and saves SRT files next to your videos.

Can I get translated subtitles without dubbing the video?

Yes. The skill creates a speech-to-text project only (no voices are generated), and your agent translates the resulting SRT itself. You pay speech-to-text rates, and the translation step doesn’t consume Perso credits, so no dubbing fee is involved.

Can ChatGPT or Claude translate video subtitles?

Not from a raw video. Language models can’t hear audio in a video file. Paired with the Perso Dubbing skill, Perso’s speech recognition turns the video into subtitles first, and the agent then translates that SRT into any language you ask for.

Which video formats are supported?

The skill accepts mp4, mov, webm, mp3, wav, and m4a files. It also takes URLs (YouTube, TikTok, Google Drive, or any direct video link) for your own content.

Is there a video length limit?

No hard cap. The skill splits long videos losslessly, processes each part, and merges the results, so multi-hour recordings work fine as long as your credits cover them. The web editor, by comparison, caps a single video at 30 minutes.

Can I translate subtitles for a whole folder of videos at once?

Yes. Batch processing is the point of the skill. Give it a folder and it searches recursively, queues every video and language combination, retries automatically every 5 minutes when the queue is full, and resumes interrupted jobs from checkpoints without double billing.

How much does subtitle translation cost?

Extraction is billed at speech-to-text rates; translation runs inside your agent’s own model and doesn’t consume Perso credits. There’s a free trial with no credit card required, so you can test it on a real video first.

What languages can subtitles be translated into?

Speech recognition covers 100 source languages, and your agent translates the SRT into virtually any language you request. If you also want dubbed audio, Perso Dubbing generates voices in 99+ languages from the same workflow.

How do I translate an SRT file I already have?

Your agent can translate an existing SRT file directly, and that step doesn’t touch Perso at all, so it costs no credits. The skill comes in when you start from video or audio: speech recognition turns the media into subtitles first.

Your subtitles, translated where you already work

Your subtitles, translated where you already work

Your subtitles, translated where you already work

One instruction to your agent. A folder of videos in, translated SRT files out, at speech-to-text rates.