0:00
publish a book and I want to create a
0:04
voice reading my whole book. It can work
0:07
for any kind of a long content. You can
0:09
create audio for anything as long as you
0:12
want with this tool called TTS free AI
0:16
and here we will have a tour of this
0:19
amazing tool. If you have created your
0:22
book, all you have to do is to create a
0:26
name him name it however you want and
0:29
um and load it. So here I have already
0:32
created a project. I have pasted my
0:35
whole book. So here I have my whole book
0:40
paragraphs and with this tool you can
0:43
easily generate a voice over. It will
0:46
use some computing power. So depending
0:47
on your computer, it will be faster or
0:51
So once you have pasted
0:53
your text, for each paragraph, you can
0:55
assign a specific voice. So here for
0:58
example, I have one voice by default for
1:00
the whole book. For the whole
1:02
conversation, what you can do is to
1:04
switch the voice for each paragraph,
1:08
each part of your text or your book.
1:13
free AI models that have to be
1:15
downloaded on your computer and we'll
1:16
have a look soon. There are different
1:19
uh various um models available with
1:21
various functionalities. So here we have
1:23
all the voices available. So far I have
1:25
not yet cloned my my voice because
1:31
uh an upper model that I'm downloading
1:34
right now, but you have all these voices
1:35
you can you can choose from. Once you
1:38
have your voices available,
1:40
all you have to do is to click here on
1:41
generate file. It will generate
1:43
paragraph after paragraph
1:46
of this audio and once it's done, you
1:48
can read it. So here, let's have a look.
1:50
Let's try to read it.
1:53
>> Chapter one. The exhibition hums with
1:55
the kind of polite noise I know by
1:57
heart. Glasses clinking.
2:00
>> So that's the kind of thing you can do
2:02
here we just uh listen to three
2:03
paragraphs but it generates the audio
2:08
And the text can be as long as you want,
2:11
limitless in terms of uh length. Uh so
2:15
you can paste your text, select the
2:16
voice, and depending on the models you
2:20
you can clone your voice. So to clone
2:22
the voice, you have to use one of these
2:24
two models available. Now in total four
2:27
voice models. I am here downloading one
2:29
model that can clone my voice. I will do
2:32
that uh in another video, but I have
2:34
already downloaded uh one model that is
2:37
good quality enough. We just listened to
2:39
it for a part of the book. It's ideal
2:44
And you can even add Chinese and
2:48
Uh it has a very fast synthesis
2:50
exceeding real-time performance. The
2:51
model is Kokoro uh 82M. You can of
2:55
course also use Windows generating voice
2:57
generation model but it's low quality.
2:59
It's actually extremely fast synthesis,
3:02
significantly faster than real-time. If
3:04
we have a look at the detail, it takes
3:06
around 0.1 second per 1 second audio and
3:10
it uses very low memory.
3:12
It can do text-to-speech but it cannot
3:13
do voice design or voice fine-tuning.
3:16
If we have a look at the next model that
3:18
I already use, Kokoro, uh it has
3:21
text-to-speech but the same, no voice
3:22
design, no voice fine-tuning, and it
3:25
generates around 1 second uh
3:28
audio per 1 second of uh CPU usage.
3:32
And then the next model I am downloading
3:34
right now uh on which I will be able to
3:36
do voice cloning, Quan 3 TTS.
3:41
It can do text-to-speech, it can do
3:43
voice voice fine-tuning, but it cannot
3:46
do voice design, and it will generate
3:48
around 1 second of audio every 3 seconds
3:51
on my computer. And finally, the last
3:54
model available, Cwen Free TTS, has all
3:58
possibilities available, text-to-speech,
4:00
voice design, voice fine-tuning, and it
4:02
will take 6 seconds to generate 1 second
4:07
So, this tool is pretty complete, and
4:11
Link in the description, and I recommend
4:14
you to try it for yourself. You can
4:17
generate some audio for free with
4:21
with the limited tool, and you can
4:23
upgrade to be able to use all the
4:25
functionalities. So, generate your um
4:30
using this tool. Link in description.
