Free Live CaptionAbout简体中文

How to Transcribe in Descript: Languages, Free Plan and Export

Add a file to Descript's script to transcribe it, fix the language and wrong words, export text or subtitles, and see what the Free plan's hour of media covers.

Summary

  • Descript transcribes a file when you add it to the script. A file used only as a layer isn't transcribed.
  • It transcribes 26 languages written in the Latin alphabet, one per file. Chinese, Japanese and Russian aren't supported.
  • Fix words with Correct, which leaves the audio alone. Deleting text in the script cuts the audio too.
  • The Free plan includes 60 media minutes a month, counted when you upload or record, whether or not you transcribe.

We make Vocena, a live caption app. This guide is based on Descript's help center and pricing page, and on the other apps' own pages linked below, all read on October 1, 2026. We didn't transcribe a file in Descript for it.

On this page

Check the language first

Descript's language list has 26 languages, and every one is written in the Latin alphabet. They include English, Spanish, French, German, Brazilian Portuguese, Italian, Dutch, Polish and Turkish. Chinese, Japanese and Russian aren't supported. Each file can contain only one supported language; files with several spoken languages aren't supported.

For Chinese or Japanese, use a different app. MacWhisper's Gumroad page lists Chinese and Japanese among its supported languages, and our MacWhisper setup guide covers it.

If you work in more than one supported language, the language list says to set a preferred transcription language or turn on Always ask before transcribing in App Settings. English (UK) has to be picked by hand, because automatic detection doesn't choose it.

Editorial illustration: following “How to Transcribe in Descript: Languages, Free Plan and Export” with an unbranded phone, laptop, or paper note chosen to fit the scene in a quiet home listening corner or small recording room.
Illustration

Transcribe a file

  1. Upload the file to your project. It lands in Files, in the Project panel on the right. The import guide also covers Zoom cloud recordings and Google Drive.
  2. Drag the file from Files into the script editor, or select it and click Insert into script. Descript's transcription guide adds a third way: the options menu next to the file name has Transcribe file.
  3. Wait for the file to finish processing and uploading to the cloud. The transcription guide says transcription starts then; how long it takes depends on the file length, audio quality and internet connection.
  4. If several people speak, Descript's speaker detection finds the voices and then asks you to name them, playing a short sample of each so you can tell them apart.

If no transcript shows up, check that the file is in the script and not just a layer, because only script media is transcribed. Files longer than 15 hours fail, and music and lyrics aren't transcribed.

Fix mistakes without cutting audio

Descript is an editor, and the script controls the recording: delete a sentence from the script and Descript cuts that audio and video too. To fix a wrong word and nothing else, use Correct, as its correction guide explains:

  1. Double-click the word and click Correct, or press C.
  2. Type the right word. Choose Correct for this spot, or Correct All to fix it everywhere in the project.
  3. For many fixes in a row, press Option-C on a Mac (Alt-C on Windows) to enter Correct text mode, type your fixes, and press the same keys again to leave it.

The transcription guide covers other problems:

  • The transcript is in the wrong language. Open the file's options menu, choose Re-transcribe and pick the language instead of Auto.
  • Names and jargon keep coming out wrong. Add them to the transcription glossary under Tools → Transcription Glossary before you transcribe. It holds 30 terms, works for English only, and adds words you've corrected three times on its own.
  • The transcript is blank or has gaps. This can happen when the left and right channels of a stereo file are out of step. In Files, choose Separate into mono files, preview the two new files to find the clearer channel, and transcribe only that one.

Export the text

Descript's export overview puts every export behind the same button:

  1. Select Export in the top-right corner of the editor.
  2. Set Destination to Local export, which saves a file to your computer.
  3. Choose the text or subtitle file and set its options, described below.

For a document, export the transcript as DOCX, TXT, RTF, MD or HTML. You can include speaker labels and markers, and set how often timecodes appear.

For subtitles, export subtitles as SRT or VTT, with a maximum number of characters per line and lines per card. Subtitle files sit next to the video; captions burned into the picture are a separate feature.

How far 60 free media minutes go

The pricing page lists the Free plan at 60 media minutes a month, 100 AI credits one time, and local video exports up to 720p. The usage guide explains the catch: media minutes are counted when you upload or record, whether or not you transcribe, and unused minutes don't carry over. One hour-long interview uses the whole month. Transcription itself isn't metered on current plans.

Hobbyist costs $16 a person a month billed yearly or $24 monthly and includes 10 media hours a month.

If you only need the transcript

Descript's strength is editing audio and video by editing text. If you only want text from a recording, a transcription app does the job with less to learn. MacWhisper's product page says its free version transcribes audio files you drag in. Otter's import guide takes audio and video files up to 5 GB, and how to use Otter on a Mac walks through it.

Vocena can also transcribe an imported audio file as a paid feature, on iPhone and iPad with iOS or iPadOS 26 or later and on Macs with macOS 15.5 or later. With the Qwen local model it splits the text by speaker (Apple's speech recognition doesn't). It plays any line when you click it and exports text or SRT subtitles with timestamps.

Sources and checks

Device testing

We have not tested these steps on a device. This guide is based on official documentation and Vocena's product information.