How to create automatic captions in Premiere Pro

How to Create Automatic Captions in Adobe Premiere Pro

Key takeaways

• Automatic captions start with a transcript. They are two separate steps.

• Choose the dialogue track or clips tagged as dialogue to avoid transcribing music and unwanted audio.

• Check and correct the transcript before creating captions.

• Caption settings let you control line length, timing, gaps and the number of lines on screen.

• You can identify different speakers, transcribe only part of a sequence and export a plain-text transcript.

Start with the dialogue, not the footage

I have a short Irish blessing on the timeline, with a couple of video clips above it. The pictures make the sequence look nice, but they are not doing the work here. It is the spoken audio that Premiere Pro needs.

If you are following along with the exercise files, the audio is under Tourism Island → Audio, in the file called All Irish Blessing. Your own dialogue works just as well. Put it in a sequence and make a note of which audio track contains the speech.

Step 1: Open the Text panel and transcribe the sequence

Switch to the Captions workspace. If you cannot find the panel, look under Window → Text. The naming is a little easy to trip over: the workspace is Captions, but the panel you need is Text.

From the Text panel, choose to transcribe the sequence. Premiere Pro creates a written transcript first. It does not create caption clips at this stage.

Before starting, check the transcription options, especially the audio source. You can use a mix of tracks, but that is not ideal if your sequence also contains music, sound effects or audio attached to footage. In my sequence, the blessing is on audio track two, so I select that track rather than asking Premiere Pro to listen to everything.

Then run the transcription. In the version shown here, the audio is uploaded for cloud processing, so make sure you have permission to process the material that way. When it finishes, play through the sequence and compare the words with the speech. The recognition can be impressively good, even with an accent, but it still deserves a check.

Step 2: Create captions from the transcript

Once the transcript looks right, click Create Captions. For a straightforward job, the default settings are a good place to start. Premiere Pro generates timed caption clips on a track above your video.

That is the lovely bit. One moment you have spoken dialogue and a transcript; the next you have captions appearing with the blessing, beginning with the line about the road rising to meet you. No line-by-line typing.

Do give the result a proper read. Listen for words the speech recognition may have missed, and check that each caption appears at a comfortable moment. Automatic is brilliant for removing the bulk of the work, not for skipping the final review.

Choose exactly what gets transcribed

Selecting one audio track is usually enough, but sometimes that track contains a mixture of dialogue and other sounds. In that case, you can tag the clips you want as Dialogue in Essential Sound, then choose the transcription option that uses clips tagged as dialogue. Premiere Pro will focus on those clips rather than everything on the track.

If you need to change your choices after the first pass, use Retranscribe Sequence. In this version of the Text panel, that is also the practical way to replace an existing transcript; there is not a separate delete-transcript command.

The transcription settings also let you choose a supported language. Make sure it matches the speech before you run the job.

Separate different speakers

For a conversation, enable the option to recognise different speakers when you transcribe. This can separate voices even if they are recorded on the same track. The result identifies them as speakers rather than knowing their names, and you can edit those labels afterwards.

To try it, I added a second voice to the Irish blessing sequence. After retranscribing with speaker recognition enabled, I could open Edit Speakers and give the two voices names. That makes the written dialogue much easier to follow before generating captions.

There is an important permission check here. Speaker recognition involves processing people’s voices, and the option includes a privacy notice. Make sure you have the appropriate permission before uploading audio and enabling it. Availability may also depend on local rules.

Adjust how your captions appear

The Create Captionsblank caption track if you would rather enter captions manually, or generate them from the transcript you have just made.

For automatic captions, the most useful controls are:

• Caption preset and format: Choose an option suited to your intended output if you have specific delivery requirements.

• Style: Select a caption style. This is where the appearance of the captions comes into the picture.

• Maximum characters: Decide whether the text should be broken into shorter or longer lines.

• Minimum duration: Set how long a caption stays on screen.

• Gap between captions: Control whether captions follow one another closely or have a pause between them.

• Lines: Choose whether captions appear on one line or two.

If you have a specialist captioning format or stream to deliver, use the settings required for that destination. Otherwise, do not feel you have to change every control. The defaults are handy, and you can come back to the details when a particular project calls for them.

Transcribe only part of a long sequence

You do not always need the whole timeline. Set an In point with I and an Out point with O at the section you want. When you retranscribe, choose the option to work within those In and Out points. That is particularly useful when you only need captions for one passage of a longer edit.

There is also an option to merge a new transcription with an existing one, rather than replacing it. It is worth knowing it is there if you add more dialogue after your first transcription.

Export a transcript you can actually use

A transcript is useful beyond captions. You might want the dialogue as text for a website or to share with someone else on the project. Watch which export option you choose, though.

Export Transcript produces a Premiere transcript file. If what you want is ordinary, readable text, choose Export to Text File instead. That gives you a .txt file you can open and use outside Premiere Pro.

The Text panel has a couple of smaller conveniences, too. You can show pauses with an ellipsis, and you can turn off auto-scrolling if you do not want the transcript to follow playback up and down the panel.

A sensible finishing workflow

1. Put the spoken audio in your sequence and identify its track.

2. Open the Text panel and transcribe the right audio source.

3. Review the transcript, including any speaker labels.

4. Create captions and adjust the layout settings if needed.

5. Play the sequence once more to check the words, breaks and timing.

That is the whole trick: get the transcript looking good first, then make the captions. Premiere Pro may move a menu or rename an option as the feature develops, but the basic path from speech to transcript to captions is the part to remember.

Key terms

Transcript

The written version of speech recognised from your sequence. It is the source used to generate automatic captions.

Caption clip

A timed piece of on-screen text on the caption track. Premiere Pro creates these clips from the transcript.

Speaker recognition

An optional transcription setting that separates different voices and assigns speaker labels, which you can edit.

In and Out points

Markers defining the section of a timeline you want to work with. You can use them to limit a transcription to part of a sequence.

FAQs

Do I need a transcript before creating automatic captions?

Yes. In this workflow, Premiere Pro transcribes the speech first, then creates timed captions from that transcript.

How do I stop Premiere Pro transcribing music or unwanted audio?

Select the audio track containing the speech, or tag the relevant clips as Dialogue in Essential Sound and transcribe only clips with that tag.

Can I export the transcript as a plain-text file?

Yes. Choose Export to Text File rather than Export Transcript if you want a standard .txt file.