Help

Install

Tarjuman is for desktop computers and works on YouTube watch pages.

After installing, pin Tarjuman’s icon to the browser toolbar: click the extensions icon (the puzzle piece), then the pin next to “Tarjuman”. You’ll copy the translation instructions from its popup.

On install, Tarjuman asks for three permissions: storage, downloads, and access to www.youtube.com pages only. A fourth, optional permission is requested only when you send a report. In Firefox there are two optional consents: one for reports, which Firefox shows in the install dialog for you to turn on or off, and one for Library thumbnails, when you turn them on. Each is explained in Privacy.

First use

1. Open a YouTube video, turn on subtitles (CC), and from ⚙ › Subtitles/CC choose the language you want to translate from. Tarjuman doesn’t send a request or switch tracks: it captures what the player loads for you as soon as you choose.
2. From Tarjuman’s buttons on the player bar, open “Download original (SRT)”. It lists every captured track with its language, type and number of segments. Then press “Download file”.
3. Open the Tarjuman popup from the browser toolbar. “From” is filled in with the file’s language as YouTube sent it. Choose “To”, then press “Copy instructions”.
4. Send the file and the instructions to the AI model you choose, and save its output as an SRT file.
5. Upload it with the “Upload translation (SRT)” button on the player bar, and the translation appears over the video. Its language comes from a code in the file name (such as talk.ja.srt) or from “To”, and you can change it before saving.
6. When you come back to the video, you’ll find your translation and its timing as you left them.

If the menu says “Turn on YouTube subtitles (CC) and choose the language you want, then open this menu.”, the player hasn’t loaded any subtitles yet. Turn on CC or choose the language, then open the menu again.

Interface language

• Tarjuman’s interface is in Arabic or English. Change it under Settings › “Display” › “Interface language”, and every open Tarjuman interface switches.
• On first run, Tarjuman picks YouTube’s interface language if it can read it, otherwise the browser’s language. Any other language starts in English.
• It doesn’t affect video or translation languages, or your files and Library. The translation instructions you copy follow the interface language.

Translating with a model you choose

• Tarjuman doesn’t translate anything itself and doesn’t connect to any model. You choose the model and send it the file yourself.
• Languages: Tarjuman accepts every track YouTube offers for the video, manual or auto-generated, and any SRT file in any language. YouTube alone decides which languages each video has, and Tarjuman doesn’t create or switch tracks.
• YouTube’s auto-translation (⚙ › Subtitles/CC › Auto-translate › the language): Accepted, labeled “auto-translated by YouTube from ‹source language›”. It doesn’t take priority over the video’s original subtitles, and if it’s the only one captured, it’s selected with that label. It’s a machine translation rather than what’s said in the video, so its errors come on top of any transcription errors.
• The instructions are copied from the extension popup with a “From” and “To” pair, for any two different languages. Their text is in the interface language at the time of copying, Arabic or English. They include the video’s title and link, and ask for a complete SRT file with the same numbers and timing, named with the requested language’s code. Your glossary terms are added if you’ve turned that on. Copying works only on a YouTube video page, because the instructions carry its title and link.
• Segment style: “Standard” keeps short segments close to YouTube’s splitting, for the tightest sync with speech. “Longer segments” merges neighboring segments into longer sentences so the model gets more context, which changes the segment count, numbering and timing boundaries.
• Very long files: Under “Splitting long files” in the download menu, split the file into parts and send each part separately. The instructions you copy afterward include a paragraph about the parts.
• If the model returns text in the chat, upload it as is: Tarjuman accepts SRT, or timestamped plain text or Markdown.
• On upload, Tarjuman compares the file with the original: “Matches the original”, or a summary of what’s missing, extra, drifted or left untranslated. It never changes anything in your file silently.

Timing

• If the translation runs ahead of the speech or behind it, adjust it with the − and + buttons on the player bar. Each press moves it by a fixed step, 0.5 seconds by default, which you can change in Settings.
• Tarjuman saves the offset with each translation, so it’s there when you come back.
• The timing menu also has “Sync by sentence…”: choose a sentence you hear and Tarjuman sets the timing from it, with “Undo” alongside.

Editor

• Open it from “Edit translation” in the icon’s popup, or from “Library”.
• Edit each segment’s text and timing, merge and split segments, search and replace, and mark what you want to review.
• It shows what’s missing from the file compared with the original, and prepares a file with the missing segments for you to send to the model and then merge what it returns.
• Saving is explicit, and edits can be undone. The file as you uploaded it stays saved.

Library

• From Tarjuman’s icon › “Library”: every video you’ve saved a translation for, with its thumbnail, title and duration, and the language pair of each translation.
• Each video holds up to three translations, one of which you make active. When you replace a translation, the previous one is saved so you can restore it.
• From there you can export the translation, copy the instructions with the video’s language pair, or delete a translation or all of the video’s data.
• Translations saved before 2.0.0 without a language show “Unknown language”. Set it from “Translation language” in the “Translation” menu, or for all of them under Settings › “Languages”.
• Video thumbnails are requested from YouTube’s servers. Turn them off under Settings › “Display” › “Video thumbnails in Library” if you want nothing requested.
• In Firefox, thumbnails stay off until you turn on that same switch, and Firefox then asks for your “Browsing activity” consent. You can withdraw it from Firefox’s add-ons page.

Moving your Library

• Tarjuman’s Library lives in the browser where you installed it, and isn’t synced.
• A backup moves it to another browser, or to a new copy of Tarjuman:
  1. In the copy that has your Library: Settings › “Backup” › “Create a backup”. A JSON file is saved.
  2. At the destination: Settings › “Backup” › “Restore from a file…”, choose the file, and review the preview.
• If you tried Tarjuman from a folder before it was published, the store version is a separate extension that starts with an empty Library. Move your Library over with a backup, then remove the old copy once you’ve confirmed it arrived. Removing it before that deletes its Library.
• Before removing Tarjuman, make a backup if you want to keep your Library, because removal deletes it.

Reporting a problem

• Where: Settings › “About” › “Report a problem”.
• When it’s sent: Only when you choose the type (bug, suggestion or other), write what happened, review what will be sent, then press “Send report”.
• What it contains:
  • Your description as you wrote it, up to 2000 characters.
  • The Tarjuman version, the browser and its version, the build package, the operating system, and the interface language.
  • You see all of this before sending.
• Not attached automatically: Translation text, a video’s ID, title or link, or anything from your Library. Your description is sent as you wrote it, and the browser adds to the request what it adds to any request, which in Firefox includes a random ID for your Tarjuman installation (details are on the Privacy page).
• Permission: On the first send, the browser asks your permission to connect to the reports server only. If you decline, nothing is sent.
• Where it goes: A private reports channel for Tarjuman’s developer, which isn’t published.
• Server limits: Five reports per hour per IP address, and one hundred per day across all apps combined. If you go over them, it tells you when to try again.
• The number: You’ll see a report number, which is your only reference. We don’t ask for an email address, and you won’t receive a reply. Keep the number if you want to mention it later.
• If sending fails: “Try again” when that helps. “Copy report” copies it as text without a network connection, so you can send it to the support email under “Contact”.
• Details are in Privacy.

Troubleshooting

Where Tarjuman doesn’t work: Shorts, live streams in progress, players embedded in other sites, YouTube Music, phones, and picture-in-picture windows.

Contact

Support email

iSultanBy@gmail.com

Privacy

Privacy policy

Publisher

BySultan — bysltan.com