Record a training script in Descript to create a custom AI speaker that converts your text to speech. Include your microphone and location in the speaker name, read naturally, and the quality of the AI voice depends on how clean your recording is.
In the Descript Drive, click the “AI speakers” option on the left side.
About 15–20 seconds.
There's a feature in descript that allows you to turn your voice into an AI voice. So you can feed it a script and it'll automatically read it using what sounds like your voice. To do that, you need to come into your Descript drive. Click on this thing that says AI speakers on the left side. And if you've already trained speakers before, as I have, you'll see them listed here. If you haven't, there'll be a new speaker button in the middle or up here in the top right, and you can type in. I recommend be very specific with what these are because you want to match your recording environment to the voice that you use, the AI speaker that you use. So you can see, my previous ones are called like ROS nt, usb Home Office. So I have the microphone that I recorded it with and I have the location that I recorded it in. So some people record in their home office and they also record at their office office or at school or whatever it might be. And those different environments will sound different. So you need to have a setting for each one in each microphone. So on this one, I'm going to call it Ross, and I'm using this little caller mic, it's called a Hollyland. And I'm just going to call it Home Office. And that's just my personal convention. But you can do whatever you want, be as descriptive as possible. Hit confirm. And what's going to happen now is it takes you to this script. And this script does two things. It trains descript on all the sounds that it needs to recreate your voice. So it's going to match your accent, your intonation, your pacing, all that kind of stuff. And it also gives authorization to descript to use your voice. I'm not going to go into their privacy policy like what they do with your data, but you can read that for yourself on their website, as far as how they store it and train with it and stuff like that. But basically you're going to read this, select your microphone, select the microphone that you're always going to use, select it from that list. And then my recommendations is read this at a natural pace, the way that you normally speak, read it with a normal intonation, and know that the sounds you make are the sounds that you're training it on. So the quality of the output is directly related to the quality of your input here. And so what you're going to do is hit that record button, read it as soon as you're done, hit stop. Don't add any additional words, just read exactly what's there, and then I'll meet you when I'm done. I'm going to go ahead and do this. I want descript to create an artificial version of my voice, ask questions and share jokes. And as soon as I hit stop, I get this secondary dialog box and it's telling me how long my recording was. I can rerecord it. Well, first of all, I can play it with this play button and see how it sounds. And if it doesn't sound right to me, it's not going to train the AI right. So I recommend do it again with this rerecord button and you'll go through that script one more time. But if you like how it sounds, hit Submit and you'll get this authorizing speaker in the bottom right corner. And it already popped up on my list here. It's going to take just a moment and we should have a success message. And there it is. Speaker authorized. You can now use it for speech generation. And that only took about 15, 20 seconds. If I go to projects, I'm going to go to this one called Sample Script. I already wrote out some words ahead of time. And you could take any script, you could take it off of a word doc and then paste it into this Descript project. And when you do that, it's going to appear as this light blue text, as opposed to when it's actual, your actual voice or an actual video, it'll be black text. But it's when it's just text, there's nothing else there. It's going to be this light blue color. And to turn this into our AI voice, this is the crucial step, this Add speaker thing right here, we have to tell descript what voice we want to use, what voice we want to turn this into. So I'll hit Add Speaker. And by the way, in addition to the ones I've trained, you can browse stock speakers, and they have all these different ones that sound like, you know, radio voices, like very, very good, professional sounding voices. But back to Add Speaker. Add AI speaker to Project. These are the voices I've trained it on. And this is the one that I just did, the Hollyland Home Office. So I'll click on that and you can see it's now labeled. It also assigns it a color. It assigned me purple, but yours may be different. And as soon as I did that, you can see those, that text is pulsing dark blue to gray. And that's just letting me know that it's processing. And each of these paragraphs will finish at their own pace. This first paragraph's the shortest, so that'll probably be the quickest. Okay. And you can see that that top paragraph is now a solid dark blue. That's letting me know it's done. And so if I click up there and hit the play button. Let's see how this sounds. Welcome to our introduction to the descript AI speaker, a revolution in audio and video editing. If you want to see how I translated myself into other languages using descript, check out this video next.
