For now, AI Speakers are only available in English.
Descript's AI Speakers allow you to create an ultra-realistic clone of your voice, so you can generate audio of your own voice just by typing. You can use AI Speakers to overdub your audio without re-recording. You can even use AI Speakers to create text-to-speech from scratch in your projects. This article will guide you in creating a new AI Speaker.
From inside a project
Make sure you’re connected to the internet.
The best way to create an AI Speaker is from directly within a project. Here, you’ll be taking a speaker label—which helps distinguish between speakers in your script—and converting it into an AI Speaker. They look the same in a project, but AI Speakers allow you to generate speech audio.
- Open or create a project.
- Add a speaker label by typing
@
in your script, or click Add speaker if you’ve already written some of your script (or imported and transcribed a recording). - Name your speaker and select Create speaker in the dropdown menu.
- Click on your new speaker and choose
...
> Enable speech generation. You’ll now see the training and authorization script.
You can also create an AI Speaker by clicking Enable speech generation next to a speaker label. This icon appears when:
- You’ve added a speaker label and written some script.
- You’ve added a speaker label and tried overdubbing your script audio or video.
- Choose your microphone and click Record to start.
- Read the full script. If you make a few mistakes while recording or you need to re-read a sentence, that’s okay. Here are a couple tips for creating a high quality AI speaker:
- Vary your expression and tone as you read the script.
- Record in in your regular recording environment with your most-used microphone and setup.
- Click the record button again to stop recording. You can re-listen to your recording and re-record yourself as many times as you’d like.
- Once you’re satisfied, press Submit. You’ll see within minutes a confirmation message that your AI Speaker is ready for use.
If a collaborator isn’t available to record directly in the app, have them send you a recording of them reading the consent statement. Then, at the bottom of the train and authorize speaker panel, select Other ways to authorize > Upload a file > Choose a file… and select the recording file. Then, jump to Step 8 above.
From your Drive view
- Open your Drive view on the desktop app and select AI Speakers in the left sidebar.
- Select New speaker, name the AI speaker, and click Confirm. You’ll now see the training and authorization script.
- Choose your microphone and click Record to start.
- Read the full script. If you make a few mistakes while recording or you need to re-read a sentence, that’s okay. Here are a couple tips for creating a high quality AI speaker:
- Vary your expression and tone as you read the script.
- Record in in your regular recording environment with your most-used microphone and setup.
- Click the record button again to stop recording. You can re-listen to your recording and re-record yourself as many times as you’d like.
- Once you’re satisfied, press Submit. In a few minutes, you’ll see a confirmation message that your AI Speaker is ready for use!
If a collaborator isn’t available to record directly in the app, have them send you a recording of them reading the consent statement. Then, at the bottom of the Train and authorize speaker window, select Other ways to authorize > Upload a file > Choose a file… and select the recording file. Then, jump to Step 6 above.