Generovanie video titulkov a titulkov offline

Here is the actual sequence, without the schema markup.

That is the whole workflow, and it is a desktop one. Yaps Studio runs on macOS, Windows, and Linux. If you already have Yaps installed for dictation, this is the same app.

If you have never opened one, an SRT file is easy to misjudge as something technical. It is a plain text file, and it is not complicated once you have seen one.

Each block in the file has three parts: a number counting up from one, a pair of timecodes marking when a line of dialogue starts and ends, and the text itself. A timecode is just a clock reading, written as hours, minutes, seconds, and milliseconds, so a video player knows the exact moment to put a given line of text on screen and the exact moment to take it off again. String enough of those blocks together and you have a full subtitle track that any video player, editor, or upload form that accepts subtitles can read.

The part that actually matters for accuracy is where those timecodes come from. A lot of subtitle tools guess timing by dividing a sentence evenly across its rough duration, which is why captions on some videos drift out of sync with the speaking the longer the video runs. Yaps does not do that. The timestamps come from word-level timing captured during transcription itself, so each line lands where the words were actually spoken rather than where a formula estimated they should be.

This is the part of the workflow most subtitle tools do not offer at all, so it is worth its own section. Yaps can translate a subtitle file after it has been generated, which means you can produce a caption track in a language your original video was not recorded in.

The order of operations is straightforward. Transcribe the video and export the SRT in the source language first, exactly as described above. Then run that SRT file, or a plain text or Markdown transcript, through Yaps' translation step, and export the result as a second SRT file in the target language. Both files sit on your machine when you are done, the original and the translated one, and you attach whichever one fits the audience watching.

A couple of practical notes on this. Some language pairs need an additional translation model downloaded before the first run, which is a one-time step and is flagged when it applies. And translation, like everything else covered here, happens on the device. A subtitle track is often the most exposed part of a video, since it is plain readable text sitting right next to the footage, so keeping the translation step local matters as much as keeping the transcription step local.

A translated subtitle track used to mean sending your footage, or at least your transcript, to a third-party service and hoping it treated the content the way you needed it treated. Doing the translation on the same machine that did the transcription removes that step entirely.

Yaps

Captioning is one of the places where "just upload it" is genuinely the wrong default, for reasons that have nothing to do with paranoia.

Footage that has not been released yet. A trailer, a product announcement, an episode ahead of its air date. The moment you upload unreleased footage to a third-party service for captioning, you have handed a copy of it to a server you do not control, days or weeks before anyone was supposed to see it.

Client work under NDA. Agencies and freelance editors caption footage that belongs to someone else, often under a contract that says exactly where the file is allowed to go. A cloud captioning tool is a fourth party that contract never mentioned.

Interview subjects who agreed to be filmed, not to be uploaded. Someone consenting to appear in your video is not the same as consenting to have their face and voice sent to a captioning company's servers. Local processing keeps the file within the boundary the person actually agreed to.

Files too large to upload quickly. Raw video is heavy. A long interview or an hour of B-roll can turn into a multi-hour upload before a cloud tool even starts working, especially on a slow connection. Processing on the same machine the file is already sitting on skips the upload entirely, so the limiting factor becomes your computer's speed rather than your internet connection's.

None of this requires distrusting any particular company. It is simply that a file that never leaves your machine cannot be exposed by a server breach, cannot be scraped for training data, and cannot sit in a retention window you did not agree to.

Being direct about the edges of this, because a captioning guide that only lists strengths is not useful.

On speaker labelling: it is genuinely useful for an interview or panel, because the resulting subtitle text can show who is speaking rather than one undifferentiated block. It is available on macOS and Windows. It is not available on Linux yet, so a Linux transcript comes back without speaker labels attached.

On mobile: the iPhone and iPad app can transcribe an imported audio or video file too, without speaker identification, which is useful for a quick text version of something you recorded on your phone. Subtitle and SRT export is a desktop workflow through Yaps Studio.

If your job is a plain transcript rather than a synced subtitle file, the deeper walkthrough on getting text out of a video, including format support and long-file handling, is in video to transcript offline.

Can Yaps generate subtitles from a YouTube link?

No. Yaps works on a video file you already have on your machine, not on a URL. If you have downloaded a copy of a video you have the rights to caption, you can import that file, but Yaps itself does not fetch video from YouTube or anywhere else.

What is the difference between SRT and VTT?

Both are plain text subtitle formats carrying the same basic idea: numbered blocks of text paired with start and end timecodes. SRT is the older and more widely accepted format, readable by nearly every video editor and platform. VTT is a newer format built for web video players and supports a few extra styling options SRT does not. Yaps exports SRT, which covers the large majority of editors and upload forms you are likely to use.

What video formats can I import?

Yaps Studio accepts common video file types including MP4 and MOV. If your file is in an unusual container format, converting it to MP4 first with any general video tool will get it into a shape Yaps can open.

Is there a limit on video length?

There is no fixed length cap, though a longer video takes longer to process, since the whole audio track has to be transcribed. The practical limit is your machine's processing speed rather than a rule Yaps enforces.

Does this really work with no internet connection?

Yes, once the app and its speech models are installed. The transcription and the SRT export both happen locally. Translating a subtitle into a language you have not used before may need a one-time model download first, which does need a connection, but the transcription and export steps themselves do not.

Does my video get uploaded anywhere during this process?

No. The video file is read by the app on your machine and never leaves it during transcription or export. That is the entire point of the workflow described in this guide.

Can it tell me who is speaking in the subtitles?

On macOS and Windows, yes. Speaker identification labels who said what, which is useful for interviews and multi-person recordings. This feature is not currently available on Linux.

Can I get subtitles in a language different from the one spoken in the video?

Yes. Transcribe the video into a subtitle file in the original language first, then run that file through Yaps' translation step to produce a second SRT in another language. Both files are generated and stored on your own machine.

Do I need to download a model for every language pair?

Sometimes. Most common pairs work without any extra step, but some language pairs require downloading an additional translation model before the first run. Yaps flags this when it applies.

Can I generate subtitles from my phone?

You can transcribe an audio or video file on the iPhone and iPad app to get text out of it, without speaker identification. Exporting a synced SRT subtitle file is currently a desktop workflow through Yaps Studio on macOS, Windows, or Linux.

Will the captions automatically appear burned into my video?

No. Yaps produces the subtitle file itself. Burning the captions into the picture, so they are permanently part of the video frame, is done in your video editing software by importing the SRT file and rendering it in. Most editors, including free ones, support importing an SRT directly.

How accurate is the timing in the exported SRT?

The timestamps come from word-level timing captured during transcription, not from dividing a sentence evenly across an estimated duration. That means the subtitle blocks track where the words were actually spoken, including in longer videos where an estimated approach tends to drift out of sync.

Can I edit the subtitle text before exporting?

Yes. Review the transcript after transcription and correct any misheard words before you export. The corrected text carries its timing with it, so you do not need to manually adjust timecodes after fixing a word.

Is there a free way to try this without installing anything?

If you want a quick one-off result without setting up the desktop app, Yaps also offers a browser-based subtitle generator that exports SRT or VTT without uploading your file. For repeated use, translated subtitle tracks, or speaker labelling, Yaps Studio is the fuller workflow.