How to make auto captions in Premiere Pro

How do I auto-generate captions in Premiere Pro without a plugin?

Use Premiere's own Speech to Text: transcribe the sequence in the Text panel, then create captions from the transcript. It comes with the Premiere subscription at no extra cost, so try it before you pay for anything else.

1. Open the Text panel (Window › Text) and go to the Transcript tab.
2. Transcribe the sequence and set the spoken language. Adobe lists 18 transcription languages, Hindi among them.
3. Read the transcript and fix names and odd words there, before any captions exist. One fix in the transcript beats the same fix in forty captions.
4. Create captions from the transcript. Since Premiere 26.3 (June 2026) you can choose Single Word at this step, and Premiere times each word to the audio on its own.
5. Style the caption track in the Essential Graphics panel and save the look as a track style if you will use it again.

For plain subtitles that is often enough. Adobe says Speech to Text works offline from Premiere 22.2 once a language pack is installed, and captions can be translated into 27 languages. Its language list has Hindi but not Hinglish.

How do I make auto captions in Premiere Pro with mouzeclick Studio?

Open the Captions tab in mouzeclick Studio, pick a model and a language, press Transcribe, read the captions through, then put them on the timeline. It takes a few presses and a read-through, so you see every word before Premiere does.

1. In Premiere Pro, open Window › Extensions › mouzeclick Studio and click the Captions tab.
2. Under Listen to, choose the selected clips or a whole audio track. Pick a Model (Fast, Accurate or Best) and a Language: Detect it, or one of 47.
3. In Spell right, type the names it should know: the client, the brand, a guest. Tick Translate to English if you want English captions from another language.
4. Press Transcribe. The first time, the speech model downloads once. You can keep editing while it works, and Cancel stops it.
5. Read the list. Click a caption to move the playhead there; double-click it to fix the words. Words Whisper was unsure of are marked, so check those first.
6. Pick a style (Subtitles, Short, Social or Word by word) and an output (Caption track, or Styled on the video), then press the button under the list.

Whisper is the open speech model, and mouzeclick runs it on your machine, with every caption timed to the words. If you also cut pauses with the Studio's silence remover or drop retakes, do that first and caption last: captions made after the cuts line up with the new timeline. It all sits beside the other Premiere Pro tools in mouzeclick.

Which speech model should I pick: Fast, Accurate or Best?

Start with Accurate. It is the default, a 190 MB download, and the model Hinglish captions use. Try Best on a clip where Accurate keeps missing words; use Fast when you only need a rough transcript to search.

We don't publish speed or accuracy figures: both depend on your computer and your recording. Clean audio helps more than a bigger model, and for a name it keeps mishearing, Spell right does more than switching models.

What do the Subtitles, Short, Social and Word by word presets change?

They decide how the transcript is split into captions: characters per line, number of lines, words per caption and how long one caption may stay on screen. Pick the preset closest to the job, then fine-tune it.

After a preset you can still set characters per line (10 to 60), words per caption (up to 8, or any), one or two lines, letter case, and Close gaps under (up to 1.5 s), which holds a caption on screen across a short pause instead of flashing an empty frame. Three switches clean the text: Punctuation, Drop um, uh, and Drop [Music] and other sound notes.

As a starting point, use Subtitles for YouTube and client deliverables, Social or Word by word for vertical video, and one line on a phone screen.

Should captions go on Premiere's caption track or be styled on the video?

Use the caption track when someone else will edit, restyle or translate the captions later. Use Styled on the video when the captions are part of the look, as on Reels and Shorts.

• Caption track puts the words on Premiere's own caption track in the preset you chose. You edit the words in Premiere and set the font and colours in Essential Graphics, exactly as with captions Premiere made itself.
• Styled on the video: mouzeclick draws each caption as a picture in the font and look you choose and lays it on a video track above everything, timed to every word. There are 289 looks in 27 families, among them Bold & kinetic, Neon & glow, Handwritten, Karaoke, Indian & multilingual and Word by word. Words come in with Pop, Bounce, Drop in, Slide up, Zoom in, Typewriter, Wipe and more. Any font on your computer works, plus mouzeclick's own caption fonts, and Premiere needs none of them installed.

The trade-off: styled captions are pictures, not live text. To change a word, fix it in the Studio and add them again. The pictures go into their own bin, and one button takes the last styled set off the timeline. Show it on my video previews a look on the frame at your playhead before you commit. For vertical work, put the platform's safe-zone guide on first, so you can see where the app's buttons will sit over the words.

How is mouzeclick different from Premiere's built-in Speech to Text?

Premiere's tool is included, runs on Mac and Windows and translates into more languages. mouzeclick adds Hinglish, 47 transcription languages, word-timed styled looks and spelling fixes that carry over to the next transcript, on Windows only.

Where Premiere's own captions are the better pick: you edit on a Mac; you need captions translated into a language other than English; the client wants plain captions they can restyle; or you already have a caption look you like. In any of those cases, don't pay for a second caption tool.

There is also a second caption tool in the mouzeclick workspace for Premiere Pro (Windows). Its Caption studio has four modes (Captions, Word Pop, Transcript and Editor), eight on-device speech models and 40 language options. Word Pop makes one text clip per word that pops in with an eased scale and fade, and its Subtitle Editor writes a new .srt without touching the original. In After Effects, the workspace builds caption layers in the comp: see auto captions in After Effects.

Why are my auto captions in Premiere Pro wrong, and how do I fix them?

Most caption problems come from names, music, a changed timeline or the wrong language setting, and each has a fix in the panel.

• Names and brands spelled wrong. Type them in Spell right before you transcribe. After, use Find a word, Replace with and Replace all, and tick Remember this fix so every new transcript gets it by itself.
• Captions sit in the wrong place after an edit. The transcript belongs to the timeline it was made from. If you cut after transcribing, transcribe again; the Retakes tab says the same, because its cuts would land in the wrong places.
• Song lyrics or [Music] notes in the captions. Leave Drop [Music] and other sound notes on, and listen to the dialogue track only rather than the full mix.
• Um and uh everywhere. Drop um, uh removes them from the captions. To cut them from the picture too, use the Studio's Retakes tab.
• Wrong language picked up. Detect it guesses from the audio. On short clips or mixed speech, set the language yourself.
• Translate to English is greyed out. Best cannot translate, so switch to Fast or Accurate. Hinglish is never translated: it stays as it was said.

Does my audio leave my computer when mouzeclick makes captions?

No. Whisper runs on your computer, and after the speech model downloads once, your audio never leaves the machine. That matters for client footage under an NDA.

One thing to know: mouzeclick itself needs an internet connection while it runs, because your licence lives in your account. If the connection drops, the shortcuts pause and your mouse works as a normal mouse until it is back. So "on your computer" here means your audio stays put, not that the app works without a connection.

Do auto captions work on Mac?

Not yet. Auto captions, styled captions and Hinglish captions are Windows only for now. The Mac app runs the mouse side (wheels, per-app buttons and gestures) and the Studio's silence, timeline and guide tools, but not captions. On a Mac, Premiere's own Speech to Text is your route, and the Mac install guide lists what the Mac app has today.

Questions people ask

Sources

Checked on 2026-10-31. Prices and features change: each link goes to the maker's own page.

1. Adobe: Speech to Text in Premiere (languages, translation, offline use, price)
2. Adobe Community: What's New in Adobe Premiere 26.3, June 2026 (Single-Word Captions)
3. Adobe: Premiere free trial (macOS and Windows)
4. AutoCut home page (captions, 90+ presets, Mac and Windows)
5. FireCut pricing (transcription hours, 50+ caption languages)
6. FireCut home page (internet connection, host apps)

Keep reading