Automatic transcription with Descript

Descript offers both Automatic (AI-powered) and White Glove (human-powered) transcription.

For automatic transcription, Descript (and most automatic transcription services) uses a third-party transcription engine — the accuracy provided by bigger companies is far superior to anything that could be built independently. We regularly measure the accuracy of all the major transcription services so we can be sure we're always providing the highest level of accuracy to our customers. We sometimes publish the results of these tests -  you can read more about these findings here.

Automatic transcription quality varies wildly depending on the nature of the audio. With a high-quality recording, and a native English person speaking clearly, accuracy can get up to 95% — but poor audio quality and hard-to-distinguish speech can reduce accuracy to 80% or lower.

Here's what we're doing to help:

  1. We offer a free trial with three hours of transcription so you can make sure it works for you before spending money.
  2. We've built powerful tools that make it very fast to correct transcripts - so you can close the gap between an automatic transcript and perfection in far less time than it would take to transcribe something by hand.
  3. We offer a human-powered White Glove transcription service to all paid subscribers. If Descript's best-in-class automatic transcription isn't cutting it for you and you don't want to go to the trouble to correct typos, then you need human transcription, and we've got you covered. Our White Glove transcription is quality-guaranteed, and costs $2 per minute, with an average turnaround of 24-hours. 

 

Transcribing a file

To get started, drag an audio or video file into a blank Composition, or create a brand new audio recording or screen recording from scratch. Once your file has been added to the Composition, it will automatically begin transcribing.
import_file_longerfade.gif

 

While the file transcribes, a window will appear prompting you to add Speaker Labels for the files you've added. If there is more than one speaker on a single file, click Enter Speaker Name, hover over Detect speakers and then select the correct number of speakers to start Speaker Detective.

detect_speakers_75per.gif

 

Multitrack Recordings

If you import multiple files of the same length at once, you'll have the option to create a multitrack Sequence.

Multitrack_Sequence_Option.png

Initiating White Glove or Importing a Transcript

If you would prefer human transcription or already have a transcript that you would like to sync, you can drag the file(s) into the Project Sidebar and select More options at the bottom. Learn more about these options by visiting the articles on White Glove transcription and importing and syncing an existing transcript.

 

Transcribing additional files on the Timeline

It's important to note that the first file(s) you add to the Script Track in your Composition will  immediately initiate Automatic Transcription. If you would like to add even more files to your script track / timeline, you will first be prompted to Transcribe in a prompt at the bottom of the Script Editor.

Transcribe-additional-files.png

 

Clicking the Transcribe button will begin the transcription process and the same confirmation window will appear. If you are adding multiple files, you will also have the option to create a multitrack Sequence.

transcribe-files-confirmation.png

 

You can also initiate transcription later on by clicking on the source file in the Project Sidebar which will open the Clip Inspector. There you can select Transcribe file to begin the transcription.

transcribe-clip-inspector.png