Synthesizing Voices with Artificial Intelligence
Hey, this is Joel Comm. But not really. It’s my voice, but what you are listening to are synthesized speech patterns that I never actually uttered. Using AI, I am able to put words in my own mouth.
Travis: Yeah, and he is able to put words in my mouth, too. Get out of my mouth, Joel.
Joel: It’s almost like I can make Travis say whatever I want him to say.
Travis: Joel is the best. Joel for President. Hey, stop it!
Joel: Today we’ll show you how you can do this as well using an incredible tool called Eleven Labs.
Travis: Joel is a rock star. He’s the best ever.
Joel: Thanks Travis. I’ll let you speak for yourself as we bring the world episode number five of the BAD AI show.
Relevant Links
Full Transcript and Time Stamps
00:00:00
S1: Hey, this is Joel. But not really. It’s my voice. But what you are listening to are synthesized speech patterns that I never actually uttered using AI. I am able to put words in my own mouth.
00:00:11
S2: Yeah. And he is able to put words in my mouth to get out of my mouth. Joel It’s almost.
00:00:15
S1: Like I can make Travis say whatever I want him to say. Joel is the.
00:00:18
S2: Best Joel for president.
00:00:20
S1: Today. We’ll show you how you can do this as well, using an incredible tool called 11 Labs.
00:00:25
S2: Hey, stop it. Joel is a rock star. He’s the best ever.
00:00:28
S1: Thanks, Travis. I’ll let you speak for yourself as we bring the world episode number five of the bad AI show.
00:00:37
S3: Three, two, one.
00:00:40
S1: Who’s bad and it’s the bad AI show. Episode number five Real Joel with the real voice here and real. Travis Right, With the real voice there.
00:00:53
S4: Yeah, it’s pretty wild. That thing is it’s 11 Labs is almost like you got a puppet up someone’s ass and you’re saying whatever they want them to say. So I’d really appreciate it. Joel, if I remove your hand, it’s uncomfortable.
00:01:05
S2: I had to go wash up.
00:01:06
S1: I use.
00:01:06
S2: Gloves.
00:01:07
S4: Just like this. Joel.
00:01:10
S5: Like. Hey, Grover. Hey. That’s. That’s why he talked like that.
00:01:17
S1: This is really amazing stuff. It’s. It’s intriguing. It’s very useful. And at the same time, it’s very creepy. Uh, 11 Labs, Dot IO is the website. I’m going to go ahead and pull it up for those of you watching the video so you can see for yourself. Beta dot 11 labs.io is the site the most realistic and versatile AI speech software ever. They say 11 brings the most compelling, rich and lifelike voices to creators and publishers seeking the ultimate tools for storytelling. And it says Here, Trev, get tell me what to type in here. I’m going to go ahead and say.
00:01:58
S4: This before we jump up in that I would say there are a few of these text to to audio tools out there already. That’s coming up. There was one what’re speech or something like that that we were playing with originally.
00:02:10
S1: It was resemble.
00:02:12
S5: Resemble. Yeah. There’s a couple.
00:02:14
S4: Of other ones that I’ve seen too, but they’re not, they don’t have the inflection down quite like this does. So this is, it’s pretty interesting. Joel’s actually typing this stuff down while I’m I’m saying this, but this technology is getting better and better. You know how with some PDFs you can just tell it to read it to you? Oh, I do it in an English accent of an English woman. Well, now you you can say have Joel read it for you.
00:02:42
S1: And so I just typed in. This is they give you a test here for free. I have the paid version I’m going to sign in a moment. We’re actually going to synthesize a voice for you. But I just typed in stuff here and you could see there’s pre-made Adam, Anthony, Arnold, Bella, dummy. Ellie. Josh, let’s go with Josh here.
00:03:00
S4: How come there’s not a pre-made Eve? There’s a pre-made Adam or there’s not a pre-made Eve.
00:03:05
S6: They don’t have the inflection down like this does yet. Joel is typing stuff while I’m saying it, the technology is getting better and better. Wow. Pretty cool stuff.
00:03:14
S1: That’s a pre-made voice that’s pre-made. Josh But the real fun comes in when you can actually synthesize voices that you’re going to use that you know that are your own. Now, if you’re making like a marketing video, like you’re talking about, you can use somebody else’s voice, right? No, no problem.
00:03:32
S4: Yeah. And some of you guys might have seen and maybe we can pull up this video later on where you could hear former presidents playing Minecraft or playing Call of Duty. And they’ve used apps like this to be able to make Obama’s voice and George Bush’s voice and Trump’s voice and Biden’s voice. And they’re playing against one another. And it’s hilarious. And they’re using a tool like this.
00:03:56
S1: Now, you do have to be careful because there are copyright rules that apply here. And when you upload a voice to synthesize, 11 labs will ask you, do you have permission to do this? I don’t think that it is ethical nor legal to use the voice to synthesize of an somebody else who, you know, has recorded voice. We’ve talked.
00:04:24
S4: Didn’t give Joe permission to use my voice earlier on and he did it.
00:04:27
S5: Anyway, no.
00:04:28
S1: Permission. So I just pulled your blog, your bio in right here, and I’m telling you to do it in your voice. I’m going to click Generate and listen into this. It’s going to happen real fast.
00:04:39
S2: Travis Wright is a venture capitalist, public speaker, digital disruptor, marketing provocateur, stand up comedian, tech blogger and brand awesome. This Travis right guy shares marketing resources, industry analysis, interweb strategy and random hilarity along w slash occasional sport or political rants. If you can stand his jokes and want his opinion on something, feel free to connect with him.
00:05:02
S5: That’s not bad.
00:05:04
S4: It did really good with made up bullshit words that I created on the fly.
00:05:07
S5: It did.
00:05:08
S1: It said awesome iser well and.
00:05:10
S5: Strategically and.
00:05:12
S1: Strategically did not. When it got to random hilarity along w slash occasional sport it said yeah instead of an.
00:05:19
S4: Understand that as.
00:05:20
S5: With yeah so so check.
00:05:21
S1: This out we’re going to synthesize a voice right now. Okay I’m going to add a voice and the voice that I’m going to add via instant voice. Cloning is my son Zach. I’ve pulled this from an episode that we did for the nifty show talking about utility and nfts and stuff. You could hear it right here.
00:05:42
S7: You also don’t expect it to eventually spike ten X or 100 x in value so that the utility that you actually wanted out of it, which was to.
00:05:52
S1: Now notice it says here.
00:05:54
S5: This is how long.
00:05:55
S4: Was that video that you uploaded?
00:05:56
S5: This is about.
00:05:57
S1: A three minute audio clip of just his voice. I edited myself out, so it’s only listening to him. It says, Providing more than five minutes of audio total brings little improvement. So really, the sample quality is more important than how long it is. And then you have to confirm right here that you have all necessary rights and consents. I actually haven’t asked that yet, but Zach, I’m using your voice. So there’s so there’s that.
00:06:22
S5: Very.
00:06:23
S4: Important to get permission and to get consent and but didn’t get any of that. So never mind. Let’s go anyway.
00:06:28
S5: Yeah. So it’s in there.
00:06:29
S1: And I’m going to put Zach in here and I’m going to add his voice. So now it’s going to do whatever it does on the back end and go, okay, now we have a voice. Look at here. It’s uploaded and ready to use now, Trav. So let’s see. I’m going to click use. All right.
00:06:50
S5: So how long.
00:06:51
S4: Was the sample that you that you did with you Because you got three samples of you and were you then you go in or you could actually read it live and you could hear it or upload.
00:06:58
S5: Something, right? I uploaded.
00:07:00
S1: It. That was a that was resemble that. I read it live, by the way 11 labs is blowing away resemble. So it still doesn’t have the inflection and the emphasis for me but I’ve realized that my voice is more I do less modulation up and down. You do more and it’s picking up on that. So cloned Zach is already in here. Let’s have clone Zach read your bio now I’ll replace w slash with with and I’m going to click generate and let’s see what it does the first time. Not heard this before now since it’s the first time cloning using his voice it.
00:07:39
S7: Travis Wright is a venture catalyst public speaker digital disruptor marketing provocateur stand up comedian, tech blogger and brand awesome iser this Travis right guy shares marketing resources, industry analysis, interweb strategy and random hilarity along with occasional sport or political rants. If you can stand his jokes and want his opinion on something, feel free to connect with him.
00:08:04
S5: I mean, that’s mind boggling.
00:08:06
S4: That sounded like him pretty much.
00:08:08
S1: It did mean Sorry, Zach, but you know, you’re a broadcaster and you’re my son, so you’re not going to sue me, dude.
00:08:15
S4: So if one. If one of us die, then. But we can still keep doing the podcast. All we got to do is say what he was going to say, plug it in, and then you can still have Travis going on into the year 2030 when we own nothing and we’ll be happier than ever.
00:08:28
S1: That’s pretty much it. So the subscriptions work like this right here. The starter plan is $5 a month. And let’s see what it tell you what it gets you here. It’s a number of minutes, the starter. Okay. So a number of characters. So I’ve used about half my characters for five bucks.
00:08:48
S4: You use $2.50 worth. Now those don’t roll over, right? Some some tools that we use, the credits will roll over. This one doesn’t look like it rolls over.
00:08:57
S1: I don’t think they do. So you could see it right here. You can use 10,000 characters a month for for free. But you have to give attribution to 11 labs. And then you’ve got the creator plan for $22 a month, which is about two hours of audio, and then independent publishers for $99 a month, that’s ten hours of generator. So basically for $99, you could take the copy of your book and turn it into an audiobook. That’s basically what you would do to create it. It turn it into an audiobook.
00:09:29
S4: Wow. And could you imagine being able to turn your audio book into your regular book, into an audiobook for 100 bucks and then sell a copy for ten bucks each or something? Right. And I was like, How many audiobooks could you sell? You know, I have a lot of PDFs, Joel, that I’ve collected over the years that I was just like, Man, it would be really nice to be able to listen to this because what I personally like to do when I am consuming media is I prefer to look at it and hear it at the same time, because then it’s using two of my input modalities instead of just one. And as we’ve gotten older and our attention span is they say less than a goldfish at this point, it’s a lot better to be able to see it and hear it at the same time. So I like that. It’d be even better if I could taste my words, but I don’t know what words taste like. I do know that some words taste like soap from experience. As a kid, I.
00:10:24
S1: Want bacon tasting words.
00:10:26
S4: Bacon tasting words.
00:10:27
S5: Those words in my mouth, they taste like bacon. So there you go.
00:10:32
S4: That tastes like dove soap. And remember.
00:10:37
S5: Don’t say that, Travis. Don’t say that. I mean, that’s it. It’s straight.
00:10:41
S1: Up. It’s simple. I want to make it more complex than it is. But the complexity happens on the AI side, the tool that these people have built. And so you guys go and be creative with it for your business and and what have you, and don’t be evil with it. Don’t be like coming up with ideas to to record something that’s not you, to to fool somebody else unless it’s just like an April Fool’s joke. That’s.
00:11:07
S4: It’s just like it’s like a fair use. Like, you know, it’s a joke. You know, Obama and Trump aren’t playing Call of Duty together. However, somebody made that as a video and they put it all together and they’re calling each other names and it’s pretty funny. And they’re using all all the Gen Y lingo and gen Gen Z lingo and all the things and get off my damn lawn. And but that’s what’s funny about it. And you know that it’s not real. However, if you can see where we are now, how close is it going to be to a point where you’re not going to believe what you see in here? Because what you see in here was created by artificial intelligence. And you don’t know if what they said or or heard was real or not. This could get really crazy in the future. Joel.
00:11:51
S1: I also discovered this here is that you can go and see your history of every single thing that that you’ve created and, and you can download them right here on, uh, on this site. So I don’t know if that ever clears out or not, but everything is here. I can just hit play.
00:12:13
S4: Yeah, I know. The very first bad show we did wasn’t that we were using resemble I. No, no. I think it was the, it was the bad crypto episode where we were talking about, wow, this world is crazy. Before we started the bad AI show, I think it used resemble and you can really tell a difference between the two because the other one seems more robotic and it just seems like maybe you were drunk that day or something.
00:12:38
S5: It also had.
00:12:38
S1: Artifacts. There’s a little crackle, a little pop in it, whereas with 11 labs, it seems very smooth. So there you go, gang. 11 labs.io. We are not being paid to share this with you. There’s no affiliate link. We just think it’s badass. Thought you would like to see it too. If you like what you have seen and learned something, then hop on over to our iTunes and give us a five star review. If you do that and you screen capture the five star review that’s posted along with your Ethereum wallet address and email us at Bad Crypto podcast at gmail.com, we will send you a awesome gold bad AI show. NFT. It’s exclusive for those who support the show. There’s a lot of different ways to support the show Right now. The call is to let other people know by subscribing and submitting that five star review.
00:13:34
S4: Yeah, right now I’m supporting these two big old bitties.
00:13:40
S1: For those who are.
00:13:41
S5: Watching.
00:13:42
S1: The video, that’ll make sense. For those who are.
00:13:44
S5: Listening to audio, you’ll have.
00:13:46
S1: To imagine what the big old bitties are that Travis is supporting.
00:13:50
S5: On the next episode.
00:13:51
S4: Coins behind me that are evolving, which is actually something we’re going to be talking about in our next tool on our next episode. So if you’ve wondered how is Travis and how is Joel creating these really cool backgrounds behind them for their videos and their zooms and whatnot? Well, we’re going to show you a process on how to do that in the next one. So if you are tuning in in YouTube, you know what?
00:14:13
S2: Hopefully you are.
00:14:14
S4: If so, click that like button. You know the deal. If you enjoy it, let us know you enjoy it and then keep enjoying it by doing the things that you need to do that you know, to do that everybody tells you to do.
00:14:24
S1: We’re not going to tell you to do. And one of the things that you know to do because we’ve told you to do it, but, you know, to do it anyway is to stay bad. Oh.
