Descript – edit your videos as easily as a text document!

November 2022

Peter Nock has been a member of TechCommNZ since he started his Graduate Diploma in Information Design in 2014. He was a TechCommNZ board member for 6 years. He enjoys trying new tech (physical or software), seeing what it can do and helping others to see what is possible with it.

Peter lets us know about a video editing tool he’s been trying out lately, called Descript.

If you know me then you know I love new gadgets and software tools! I stumbled across this new tool recently and thought I’d try it out. I have my own YouTube channel, and this looks like a great tool for editing. It has a SUPER AWESOME AMAZEBALLS trick too! However, the setup for this feature takes some investment time…

Descript is a video editing tool. Ok, nothing new in that. But is has a killer feature! It can transcribe your audio (speaking) into text… ok – again, nothing new in that right? But here is where the amazeballs feature comes in – you can edit the transcribed TEXT and it will change the AUDIO in the video! OMG, right?

So, say for example the original text was “here is a green car” and you realise you had made a mistake and it was a yellow car – you edit the text to “here is a yellow car” and the audio will change to match the text – in your OWN voice! You can also add words and remove them too.

“How does this work?” I hear you ask. AI is the answer. Yes, artificial intelligence is in the background making this happen. And that is where it takes time and investment to make this work. Almost all readers will have experienced talking AI when talking to their phone (Siri, Google), their computer (Google, Cortana), their Alexa device, or when calling a company like Vodafone. The AI behind that voice can listen to human speech and match that to a database for an appropriate response. It takes a lot of training to teach the AI system to understand people’s queries and to provide the best responses. People can ask the same question in many ways.

For example:

  • How do I get a new phone?
  • I want a new phone please.
  • I need a cell phone.
  • Get me a phone now!

The AI system needs to understand the request to give the appropriate response.

Now, in the case of Descript, it is a little different. If you want to be able to edit your voice by editing the text, then you need to train the Descript AI tool on your voice. This takes between 20 to 100 hours to do – that is, you need to feed in that much of your voice audio for the AI to be able to speak for you in your voice. They do supply scripts so you can read to record your voice to train the AI.

So now your AI voice is all trained up, what else can you do? How about one-click removal of all those filler words? Um, er, hmmm etc – all gone with a single click! Made a mistake (e.g. green instead of yellow) – just edit the text. Need to add a word or delete a sentence? It will do that too! When you delete some text it seems to cut that part of the video too…

Descript is a subscription-based software-as-a-service product. They have a limited free option, and the pro option is $30 per month for all the features. They have a middle and an enterprise option too. If you are doing a lot of video editing (especially for blog-type videos) then it would be worth looking at this as your one-stop video editor.

Just to finish up, AI is getting into all sorts of software now. For example, this document in Word has an editor score of 98% - I only missed the last 2% because I don’t think e.g. needs a comma after it! But hey, who am I to argue with it?

Note: a subsequent Google search proved that I was wrong about the comma...