Descript introduces upgraded AI voice tool

New capabilities include improved text-to-speech and faster authorisation

Podcast and video editing platform Descript has launched an enhanced version of its AI voice generation tool, and plans to release a number of new AI-powered features built into the platform’s workflow before the end of the year, the company announced yesterday. 

The platform's new AI Speakers tool allows users to generate and edit artificial voices based on pre-recorded samples. Formerly known as Overdub, the functionality also makes it easier for users to create a new AI speaker and to edit it through text-to-speech technology. 

To create a new AI speaker, users need to access the drive view and click on the AI speakers tab (formerly the Voices tab) and upload an audio consent statement recorded by whoever the AI will be creating a voice clone of, using their own voice. Descript recommends that the consent statement includes a variety of different tones from the speaker to make the AI voice less monotone and robotic. 

Descript also says that the upgraded version of the AI voice tool promises “much, much better” quality and claims that users will have the ability to create an entire voiceover without having to record a single word. The approval process is also now much faster, taking less than a minute, with users no longer having to wait 24 hours to train or verify a voice. 

In addition to the new AI voice tool, Descript said that it will be introducing a whole new series of AI features built into the Descript workflow within the next six to eight weeks. The aim of the features is to help make users’ creative workflows easier, with AI acting as an assistant that can help pick up menial tasks while the creative decision-making is left to the user. 

“We didn’t want to release a bunch of AI stuff just because we could, or trot out some AI party tricks that wouldn’t help anybody do anything except produce a flood of really weird, bad content,” said Descript’s head of product Laura Burkhauser. “Instead, we wanted to be sure that Descript’s new AI features would actually be useful. At its best, we think AI can be like your junior editor.”

“It’s an assistant you can order to do the technical, tedious stuff you don’t want to do yourself and it's a collaborator that can kick ideas around with you, offer suggestions, point out flaws.” 

At the start of this month, Descript officially shut down its older Classic interface - after announcing in July that it would be phased out - and users have had to transition their older projects to the newer Storyboard interface to continue working on them. The new interface features a number of new solutions that make it easier to edit audio, video, images, and captions through a script format, and the company claimed that it fixed “over 1,000 bugs” from the old interface. 

As part of its new offerings, Descript acquired remote recording platform SquadCast in August and integrated its solution into Descript’s workflow. This allows users to record remotely in both audio and video with up to 4K quality, as well as offering automatic backups for each guest, the ability to join via mobile, and automatic transcriptions which can be edited on Descript.