Whether an amateur or even professional podcaster, all these individuals are well aware of the painstaking process of editing that takes place behind the scenes. Specifically, the copious errors behind human speech patterns and the precise editing to correct these mistakes require large amounts of time/experience to fix.
“Load up the audio. Listen. Drag. Delete. Listen again. Restore the original audio. Hone. Delete again. Re-listen. It’s exhausting. And that’s just for the little things. There’s a better way” (Laforme, p.1).
Descript has become one of the most popular and innovative editing platforms specifically for the production of podcasts. This tool creates a text transcript of audio files and as users delete or rearrange the transcript, the same tool edits the audio file to match it exactly essentially eliminating a whole area of editing required to produce clean production. Formerly, the application required users to pay for every new production they would create, but recently the company has moved towards paid monthly subscriptions for unlimited edits/projects.
The application announced over the weekend that its new version update will add a number of new autonomous podcast editing tools, but most importantly the software is also being integrated with Lyrebird, a “text-to-speech tool that the internet lost its mind over in 2017.” Podcast editors who use Descript will now have the ability to simply type text into the application and have a custom audio generated sound produced.
While this new integration with Lyrebird could prove to help editors drastically such as allowing them to fill in a missed word or two on their audio tracks, many are worried about this software that could allow users to “clone” someone else’s voice.
Misinformation has been a growing problem over the internet and news sources as the spread/access of information continues to grow increasingly more powerful every day. Podcasters using Descript have expressed significant amounts of concern towards this new update in fear that auto generated voicing could lead to editors cloning other individuals voices and releasing their productions online.
A great example of this potential danger could be unleashed in the upcoming Democratic candidate race. With leading candidates such as Biden topping the boards, this kind of auto voice generation could make it easy to clone the sound of his voice releasing false statements all over the internet. Though many can refute this concern by saying the voice generation was fake, a cloud of misinformation and productions could make it increasingly more difficult to differentiate between fact or fiction.
Descript claims to have recognized this potential threat in media and released an ethics statement on its new update saying “We are committed to modeling a responsible implementation of these technologies, unlocking the benefits of generative media while safeguarding against malicious use” (Descript).
Although still early in its development/innovation, auto voice generation software is growing increasingly more strong and clean, making it easy to mimic another entities voice. When combined with a growing media industry such as podcasting and presented with the motivation of saving time on editing, the spread of misinformation could hit the industry like a title wave. Accurate information and statements is a growing area of concern in today’s society and media experts believe applications like Descript although intending to do good, are potentially pushing this issue further.