How to Give Your AI Characters a Consistent Voice

0:04Did you send for me peace, witchcraft, idolatry? You have sucked the life out of God's people, and I am his justice. All right, so now I'm going to show you how to attach a voice to a character so that you have better character consistency throughout your videos. Now, what you do is click on your actors tab. go to your actors tab and you'll end up generating an actor or creating one manually. And I'll just show you one that I've made here for e-hood.

0:39And when you go into your actor, you will start to notice that there is a voice function and parameter that you can add to your character. And there are four different ways that you can add a voice to a character. One is from the image. And so literally what that means is Pray Production Studio will look at the image and then create a voice based off of what it sees here. So usually your headsh shot or you can go to the describe button and literally describe what the voice should sound like. You

1:10can also um click on 11 Labs and pick one of the 11 Labs voices that you can add to the character or you can upload your own. So if you end up uploading your own, just try to make sure that it's about 30 seconds long or less. those will work the best. And so from here, this is what it looks like. You can play it back and just make sure that when you're done and you make your voice, make sure that it says saved right here. And if it doesn't, just kind of scroll down and hit that save changes button so that you have the character

1:41voice locked into the character. And so once you do that, then you have your voice attached to the character. So then when you go to your project and you're on the video tab, you need to make sure that first the character is a part of your project. So on the shots tab, you'll notice that you have your um characters here and your actors attached to the characters, right? So we have E-hood here, the one that we just put the voice on. And so you can see that he's attached to the project. So then by

2:14the time you get to the video tab and you start generating video that you're able to do that. And so this is what it looks like when you go into your video generation. Just make sure you're in Seed Dance 2.0, the new version that we have. And choose Seedance 2.0 or Fast. Those are usually the best ones. If you want to even save some more credits, um mini is good as well, but I'll do uh Seedance 2.0 on this one. Just dial in your parameters here. And then this is super important right here, the text and

2:47motion prompt. Okay, so when you start to write in what your character is going to say, just make sure that you name your character in the prompt. So as you can see here, it says e-hood tilts toward the mic with dry unhurried delivery and says this line. And now you'll notice that the dialogue that he says are in these brackets. Very important. Seedance 2.0 know actually works better with these brackets than with quotes. You know, it works with quotes, but only sometimes. I've noticed that um

3:19a couple of generations don't go through. So, the brackets actually work better. And you'll see I have Eglon and what he says, too, right? And so, you can't say he straightens his spine to full seated, right? you have to say Eglon so that the generator knows that this is the voice I'm going to pull when saying this line. So that's super important. Now we do have this Pray Production Studio enhanced prompt button right here. So you can write in your own

3:49prompt and then click this button and what it will do is read the prompt that you have written in here and then expand upon it and write it in a seed dance 2.0 um way that works better for the generator. So, we did put that in there for you. And then from there, that's all you really need to know on how to attach voices to actors. And then you'll get a generation that looks like this. Welcome back to the Fat King podcast. I'm Ehoud, judge of Israel.

4:26And I am Eglon, King of Moab, conqueror of Jericho, your host. All right, great. So, what I've noticed is two characters works really well. One character works awesome, and three characters sometimes works, sometimes doesn't. Beyond three, you're going to have some issues. So, keep it to three or less. um you try to push for two and one works the best obviously. Now one

4:55more thing before I go. If you are generating with Seedance 2.0 on this generation, this was a first and last frame generation. If you choose references, okay, and you upload an audio reference and use that audio reference, it's going to override your character's voice and audio. So that's just good to know. Um, we've had some people in the past override it and wonder what happened to the character's voice. It's basically that's what

5:25happens. So, if you upload a um reference audio, it will override your character's audio. So, usually on these um character shots where you want to make sure that your character is saying the words and the phrases inside of the prompt, I usually like to do first and last frame. And if I'm doing references, then I'm not uploading a reference audio with this generation as well. So, hope this is helpful. Go out there, test it out, let us know what it looks like, how

5:57it sounds, and excited to see what you guys end up doing. Go make some epic