We got a bunch of questions from the episode where I talked about AI for voiceovers, so this week we dive into the fascinating world of AI voiceovers and how they're shaking up the L&D industry. Thanks to incredible advancements in machine learning and natural language processing, AI voices are now impressively human-like, capable of conveying emotions, tones, and all the quirks that make us unique.

We’ll explore the journey from robotic-sounding voices to today’s remarkable AI-generated speech. Meet 11 Labs, the Swiss Army knife of AI voiceovers. From seamless text-to-speech and versatile speech-to-speech features to an extensive voice library filled with diverse accents, ages, and genders, 11 Labs is your go-to for high-quality audio content.

We’ll also discuss the great debate: AI versus human actors. While AI offers speed and budget-friendliness, there's still something irreplaceable about the raw emotion of a human performance. Maybe a mix of both is the future sweet spot?

Join us as we look ahead to what’s next for AI voiceovers, creating lifelike audio experiences you won’t believe. Tune in to learn how you can harness this technology to create killer audio content without breaking the bank.

–––––––––––––––––

At ttcInnovations, we help businesses create lasting change with immersive learning experiences. Through instructional strategy, design, and content development we empower employee confidence, performance, and results.

💡 Looking for custom learning experiences without licensing fees? Contact us for a free consultation! https://bit.ly/4aOhPKq

🤝 Need extra hands fast? Try staff augmentation! Click here to get matched with experts in 48 hours - no job posting needed. https://bit.ly/4aOhPKq

🚀 Simplify outsourcing with our subscription plans - predictable pricing and limitless innovation. Book a meeting for your free first week! https://bit.ly/4aOhPKq

🎯 Boost results with serious games for optimal retention and results. Contact our Dashe team to get started! https://bit.ly/4aOhPKq

Welcome back to Learning Matters. I'm Doug Woldridge, your host. And today we're going to take a look at 11 Labs. It is a software that I like to use when working on AI voiceovers that allows us to do speech to speech and not just text to speech. So without further ado, let's get to the show. Okay. Welcome to my audio studio. Today we're going to take a look at 11 Labs and see what all options they have for AI for voiceover.

It's a pretty cool tool. I do want to do a quick disclaimer real quick that they are not sponsoring this video. I have my own subscription with them and have been using their product for a couple months now because I find it fascinating. I don't know if this is true, but I think they're one of the only AI voiceover options that actually does speech to speech, which is in my mind a massive lift in

the usability as well as added quickness to working on AI voiceover stuff. Back in the day, you used to have to sit there and really play with your script and do text to speech to try and make sure that it pronounced words correctly, also added in the right pauses and things like that and really kept your flow going and the vibe. And also just kind of like a true human element. So,

We're going to walk through that. I also want to say that, no matter what my preference is to always use true, live actors. I think it's super important to make sure that, we keep the industry alive. And if you do have the option as far as timing, and as well as budget, I would highly recommend finding some great actors in your area that you can bring into your studio to work with them. I think it.

For a few reasons, I think it's easier to work with a live person in the booth. I think it's quicker because you can immediately make those changes to whether the tone isn't right, the vibe isn't right, whether the energy level isn't correct, and it's much quicker to get what the client's looking for. And it also allows for the client or the writer to come in and sit in on the session. So whether that's Zoom or on the phone, whatever it is.

It makes it super easy to do that and it gives true life to the script because, well, you're using a live actor. So with that said, I think we should dig in and start taking a look at what 11 Labs has to offer us. So let me pull up my screen real quick and we'll play around with it.

So here we have the basic setup as soon as you jump into 11 Labs. You have their voice library, which is, if I may say so myself, incredibly extensive. So right now they have, I wanna say somewhere around 150 or 200 different options here. And you can see that.

You not only have American accents, you also have Brazilian accents, British accent, Australian, so on and so forth. So you can really open the possibilities up to diversity within your script that you're working on. A lot of the clients that I work with are Fortune 100 and Fortune 50. So they're worldwide. They want to be using different types of voices that are similar to who they're training.

which it makes sense, right? you have an audience that is, let's say in England, what you would kind of like them to hear similar voices to what they hear. So let's play around a little bit and I can kind of show you the differences between text to speech. And then we'll get into the speech to speech stuff, which is way more fun. So I have over here picked out a bunch of different options for us to look through. We probably won't go through them all. There's too many.

for this podcast and basically you start with just kind of looking through their voice library and just adding someone into your

VoiceLab and then you can go to the VoiceLab here and then start playing around. I believe you can have as many as you want on here. If you just bring in like 50 people, it's going to be tough to search through everything. So if I was working on a regular project and let's say the client gives me a call and they say, hey, we would like to get a couple of different options for a few Latin American or Spanish voices.

I'd pull in a couple of different options for them here and then I would option those out, send them the files for that and then they can choose which voice really kind of matches what they're going for. So today we're gonna start with Mary. So basically you just start here with us and then you can come in and go into text to speech or speech to speech. There's this quota option here obviously I've got plenty to work with and so I believe that's on a monthly thing too.

the quota always repopulates at the end of the month, which is pretty nice. You have a bunch of settings here, which kind of just allow you to really play around with how the voice algorithm is working and where you can have like more variability. So it's not so monotone. Obviously the more stable it is, it's probably going to function better. You're going to get less artifacts that come into play.

And then similarity would be also working towards kind of that keeping away from being that monotone type of sound. Style exaggeration, so that's more based on for the multilingual type of things, whether that's using different accents or whether it's different languages. And so I kind of like to start roughly with these settings here. I can always change them later.

But let's start with hearing how Mary sounds with just the text to speech.

Hi, my name is Sofia. In this module, you'll learn the best practices of how to prepare and pack a shipment for air travel. Sounds pretty nice. I would say using a lot of different AI tools like this in the past, their text to speech is pretty spot on. I mean, I think it rivals any of the other text to speech out there. And it's a good start. I think this is a really

fairly basic and you know passable use of this voice for this script. I think it's also just a little bit boring. I don't think that it has the life that I want when I'm listening to it or I want my learners to have when they're listening to it. You know we don't want to bore people. So let's move to speech to speech and this is where it can get pretty fun. So I'm going to record the audio here myself and then my voice is going to become

Mary's voice.

Hi, my name is Sophia, and in this module, you'll learn the best practices of how to prepare and pack a shipment for air travel.

All right, let's see how that sounds.

And it does take just a little bit longer because it's going through its process.

Hi, my name is Sophia. And in this module, you'll learn the best practices of how to prepare and pack a shipment for air travel. Okay, okay. Not my best work, but it gives you a little bit of difference to it. So I can add my own pauses. I can add my own flair to it in any actor that I bring in that, that

Any actor that I bring in that comes in and reads the script, I could also change their voice if the client wants something different. So let's try another option here. I'm going to go back to our voice lab and let's try.

TRIA.

And I just kind of created a couple of like very basic scripts for us to use today. There wasn't really, I don't want to pull anything that my clients have given me or any scripts from previous projects. and I didn't really want to use chat GPT for this cause we're already using AI. So I just put some things together. let's start with the text to speechless here. how Priya sounds.

Zara has been with the company for five years and is preparing for her sabbatical. My name is Priya and we'll go through how she can best set up her team for success while she's away for an extended period of time. Okay, okay. Sounds pretty good, right? Let's go to the speech to speech. Let's see what we can do with this.

Zara has been with the company for five years and is preparing for a sabbatical. My name is Priya and we'll go through how she can best set up her team for success while she's away for an extended period of time.

One thing I'll say while this is generating is that unfortunately I can't take an entire script and speak it into here. And it does also allow you to do an upload function. So if I wanted to, instead of just recording this on the fly, I wanted to take a bunch of audio clips that I've already gone ahead and split up, edited, all that good stuff.

Run it then who am I to argue? So let's hear how this sounds Zara has been with the company for five years and is preparing for a sabbatical My name is Priya and we'll go through how she can best set up her team for success while she's away for an extended period of time. Okay? Wonderful, so one of the things That is unfortunate about this is that you still can't quite get it to do

the accent that you may have been looking for for bringing in Priya as this voice. So I think they still have a little bit of work to do there on that side of things, but it is already a game changer for me to give this as an option to a client because I'm just using my voice here and I can already get it to sound like several different characters here. So I think it also depends on which voice you choose.

going through the voice library, not all of these are winners here. So, you know, as you would expect with a massive type of library like this, there's some really great voices, and then there's some voices that oddly just don't sound very good, but just having the options here makes life a lot easier. One of the main reasons for using this would be that, let's say, let's say,

that a client is asking me to bring on maybe four or five different voices to use that are Canadian French accents. And back in the old days, I would have to do a casting call to get a bunch of different actors that could do that type of accent, either bring them into the studio or have them recorded on their own and send me little clip -its and then send that out to the client. Well, I can just go in here into the voice lab and

immediately bring up, let's see, who do we have here?

Let's try a Canadian French. Let's see how he sounds.

What does it take to make your first successful sale of our new software platform? My name is Elliot and I'll walk you through the steps and what to keep in mind when sparking with a potential new client. To be honest, I don't know how Canadian French that sounds in my mind, but it's a good voice. I do like how the voice sounds. So let's do a speech to speech and hear how that sounds.

What does it take to make your first successful sale of our new software platform? My name is Elliot and I'll walk you through the steps and what to keep in mind when you're speaking with a potential new client.

What does it take to make your first successful sale of our new software platform? My name is Elliot and I'll walk you through the steps and what to keep in mind when you're speaking with the potential new client. Okay, okay. I can hear a little bit of the accent coming through. I think that's actually fairly passable for this. So I'm going to go ahead and pull this out of here and

Let's showcase a couple of the other tools here.

So...

Not only do we have voiceover type of things, we can also generate sound effects. Let's see how a car crash sounds.

I'm gonna be honest, I haven't played with this function of it yet. I'm really excited to kind of walk with you guys and play with it with you.

Okay, generation one, two, three, four, nice. Gives us several options. Okay.

Not bad. Let's get a little more detailed with it.

All right, let's see what that does.

Interesting.

Alright. Ummm...

Let's try this. Now I'm just having a lot of fun here, so bear with me.

Okay, okay. I don't think I'm gonna use these for like a Hollywood movie, but I think these would be totally usable for a pinch. And I think one of the other really cool things that 11 Labs allows you to do is add a generative or cloning a voice. So.

you can either take a file in here, like let's say, well, here's a good example. I just had a project that I was working on and the scripts had been delayed a little bit and my actor that I had, that I was bringing into the studio had to go on vacation. I mean, didn't have to, got to go on vacation. And so we had to kind of,

pause the recording process for about a week while we waited for her to return. And we'd already done a couple of scripts based on this particular project. So we were already working with a character that we had developed and had a voice for. So I could have brought several of those files into here to make her voice. Now, I don't think that I would do that in a normal situation. I think...

Especially if the scripts had been running a little behind. I think the client usually is like, yeah. Well, no worries. We'll get it. We'll pick it up when she gets back. But let's say we had pickups that we had to do maybe like maybe they changed the script on two or three slides that we were working on and she was out of office for let's say a couple of weeks and we had to get the project done. This might be a good option for us to come in here and drop her voice into it.

and try to match that. Now let's see how good it does. Again, have not used this functionality yet, but let's play around. Let's see what's up.

Welcome to the training course for how to wire and maintain a database. My name is Noah and I'll be taking you through this journey. Click next to continue.

Welcome to the training course for how to wire and maintain a database. My name is Noah and I'll be taking you through this journey. Okay. Click next to continue. Let's sign the paperwork here.

Young -ish male American accents.

looks like we need to have multiple samples. So I'm not going to go through this full process here, but let's just say that there are options here for you because it looks like we need 25 samples and I want to walk you guys through 25 samples of me talking about things. Plus I don't really have the script ready for you to do that. But yeah, there's either way, there's just a lot of options right now in the world of AI for voiceover that I think

are fantastic. I was a big skeptic when it came to this type of technology being used specifically for voiceovers because I felt like we would never get anywhere close to passing the uncanny valley. And I still think that there are definitely times when either artifacts are inserted or it just doesn't

white gets to that truly human voice, but I do think it's a totally usable option, especially if the budget's not there or if you just don't have the time to do a true casting call for voiceovers, actors. So I think it's really fun also just to play around with this. And, you know, when I was, when we had our episode with Eric about all things AI and image creation,

One of the things that he mentioned was just, it's fun to play around with new technology. And even someone like me who is definitely a hard noser when it comes to trying to keep the actors employed and keeping that human side of things, it is really fun to play around with these new technologies and see where we're going with them. Because I think, especially within the last two years, this has become

This has become quite a more usable product as well as something that is getting closer and closer to that human element. Will it surpass it? Probably not, but who knows? We'll see in the next couple of years where 11 Labs comes into play and how they can keep expanding their options. You know, the best case scenario for me would be that they expand their ability to take

my hodgepodge of an American accent and be able to do the speech to speech and have the voiceover or the algorithm really take on the accent that that voice was built for. So if we have Mary here, I would like it to sound a little bit more like it is truly a Spanish slash Latin female. And again, while it's pretty close,

that right now. It's not quite there yet. So I hope you guys enjoyed this little walkthrough through 11 Labs. This is kind of a different setup for us than our normal podcast, but fear not. We are going to get back and sue some more interviews here coming up in the next couple of weeks. But you know, it's fun to change things up a little bit. Thanks so much for joining us this week. As always, remember to sign up for our newsletter, The Buzz.

and like and subscribe wherever you get your podcast. See you next time.