how you can use the most realistic
sounding text-to-speech software that go currently exists. Here's an example of what it sounds like.
but he delivers solid content.
but for me personally, that sounds like an actual human read that text.
The vocal emotion, the intonation, it all just sounds so realistic.
And in fact, it probably sounds even more realistic or more human-like than I do.
Let's check out how you can use this for free right now. To start turning text into speech, head
to the Eleven Labs homepage. You can click on the card in the top right-hand corner, or you can also
click on the link down below in the description. Right here on the homepage, you don't even have to
set up an account. You can just type in some text and right here, you can select a voice
to narrate that text. Here you have all of these different pre-made voices and later on, we'll look
at how you can even create your own customized voice. You have a mix of males and also females.
Once you're ready to generate your speech, you can click on this play icon right down below.
On the homepage, you are limited by how many characters you can turn into speech.
To be able to turn more text into speech, up above, let's click on the pricing tab. This
drops us onto the pricing page and here we see all of the available plans. Now the great news is the
base plan is entirely free. So here you could test out speech synthesis to see how it works.
Right down here on the free plan, you get up to 10,000 characters per month that you
can convert into speech, and I found that that works out to about 10 minutes' worth of speech.
One of the drawbacks of the free plan is you cannot use it commercially and you also have to
attribute back to Eleven Labs. The next plan up is the starter plan and this one's attractively
priced. For the first month, it's only $1 and then it jumps up to $5, and the key difference is you
get up to 30,000 characters per month that you can convert into speech, and that works out to about
30 minutes. And one of the really great parts of this is you get instant voice cloning. What this
means is I can upload, let's say myself talking, and then I can type, and it'll speak back using
my voice. That's pretty incredible. In the top right-hand corner, I'll click on sign in, but if
you don't yet have an account, feel free to sign up for a free account. Once you finish setting up
your account, here we drop into speech synthesis and this is where we can type in some text,
and it'll generate speech based on that text. The benefit of going through an account is you get
a much larger quota to convert text into speech. Right up above, just like we saw on the homepage,
here you could select one of these voices and here you could even preview what these voices sound
like by clicking on the preview icon. You can go through and listen to all the different options.
Below that, you also have voice settings and here you could adjust things like the
stability. If you go down to the left where it's more variable, this will cause the voice
to sound more expressive. Here, you could also modify the clarity and similarity enhancement.
You could try adjusting these different values to see how that affects the voice. Down below,
here's where you can type in your text and the more text you type in, the better the model will
work, and the reason why is it adjusts the delivery based on the context of the text.
Here, for instance, I'll type in text. Unfortunately, I have some bad news to report,
so it's bad news, so the voice will probably be a little bit more downbeat,
and it'll factor that into the way it delivers the voice. It's pretty incredible. Right down below,
all I need to do is click on generate and then we could listen to the voice. Unfortunately, I
have some bad news to report. Mrs. Fields won the best cookie of the year and not the Kevin Cookie
Company. That sounded really human-like. I am still just blown away by how good this technology
has gotten. Right up above, I can change the message and maybe we go with something a little
bit more positive or upbeat. Actually, we were just kidding. The Kevin Cookie Company won. Right
up here, I can choose a different voice. Let's try Elli to see how that one sounds. Down below,
I'll click on generate. Haha, just kidding. Great news to report. The Kevin Cookie Company has the
best chocolate chip cookie. Yay, this is fantastic news. That too also sounds fantastic. If you like
the voice, but you weren't quite satisfied with the delivery, up above, you could click into
voice settings and you could try modifying these different values or you can simply go back down to
generate and you can simply regenerate, and you'll get a slightly different variation. Let's listen
again. Haha, just kidding. Great news to report. The Kevin Cookie Company has the best chocolate
chip cookie. Yay, this is fantastic news. So, there you get the same voice, but it's slightly
different and you could go through and generate again and again until you find one that matches
exactly how you want it to sound. Along with the pre-made voices, if you scroll up the page,
you also have the option to add your very own voice. You can click on this or up on top, you can
click into the voice lab. Within the voice lab, you have two different options for making a new
synthetic voice from scratch. Over here, you could design a voice, or you can also clone a voice.
Let's test this out by clicking on add voice. Let's start by looking at voice design. This
opens up a prompt where I can now generate my very own voice. At the top, I can select a gender. I'll
select male. I can also select the age. I'll go with old. Right here, I can also select the
accent. I'll select British and here I can define the accent strength. Down below, I'll type in my
text. I can confirm the Kevin Cookie Company does indeed have the best cookies in the world. And
down here, I'll click on generate. I can confirm the Kevin Cookie Company does indeed have the best
cookies in the world. How would I know? I'm old, wise, and a British accent speaks with authority.
We're going to have to use this in our next marketing campaign. Here, I'll minimize this
section on the bottom and then I can select use voice. Here, I can give my voice a name. I'll type
in a name and then click on create voice. Now, anytime that I want to synthesize text to speech,
I can use this specific voice. So, this might become the voice of our marketing campaigns,
and this way I can ensure that we always have the same exact voice. Let's now add another voice,
but this time we're going to use the instant voice cloning. Here, I'll click on add voice.
And right over here, let's select instant voice cloning. I can now add my very own voice
by uploading sample audio and what better voice to try than my very own voice. Here,
I'll type in Kevin for the voice name and here I need to drag in sample audio. The good news is I
record my voice a lot anytime I pull together videos. So, I have lots of sample audio. Here,
I’ll drag a file in. They recommend uploading at least five minutes' worth of audio, providing more
doesn't really help the model that much. Right down here, I could also apply different labels and
I need to enter a description. So, I might as well say a handsome, amazing, incredible, brilliant
young man with a beautiful sounding voice. I mean, that's an accurate description, right? Down here,
you also have to confirm that you have the rights to this voice. This is my own voice. So of course,
I have the rights to it. Down at the very bottom, I'll click on add voice. And just like that, it
has now added my voice and it went really fast. It probably only took about 10 or 15 seconds to add
my voice. Right down here, I can now use my voice. Let's test this out. This brings me back to the
speech synthesis page, and right here in the voice list, we now see two new voices. Here, I have my
voice, the Kevin voice, and then I also have that other voice, the marketing campaign voice that I
can choose from. I'll select my voice. I'll leave the speech settings as is, and then right here,
I'll type in my text, and I'll use the intro of this video to see if this AI text to speech
could potentially replace me. Down here, let's click on generate and hear what it sounds like.
Hi, everyone. Kevin here. Today, we are going to look at how you can use the most realistic text
to speech software that currently exists. Here's an example of what it sounds like. Now personally,
I don't think it sounds like me that much, but let me know what you think down below in the comments.
Did that sound like me or did it sound like a computer was speaking it? Once you've converted
text into speech, up at the very top, click into the history tab, and here you can see a history of
all of your different generated samples. You can play them back again to hear what they sound like,
and if you'd like to download any of the generated speech, you can click on this icon and then you
can download the selected speech. I am just blown away by how good text to speech has gotten. The
next time you listen to an audio book, will it be an actual human or a computer narrating the book?
And will you be able to tell the difference? I don't know.
Comments
Post a Comment