There are currently two reliable ways to clone our voice, one promised by Microsoft and the other offered by ElevenLabs.
The problem is that one of them offers a cloned system or imitation of peoplebut it is not yet possible to test it, although we can enjoy listening to several examples, while the other one will allow us to use it.
Be that as it may, it is clear that Artificial Intelligence is experiencing a boom, not only in broad strokes, but it is also seen in things as “simple” as the imitation of our and other voices, encompassing such large companies is how Microsoft can .
Let’s see the two examples that we are talking about, to know what they are like and how different they look.
VALLEY
From Microsoft they are working on a model of text to voice (TTS) that can imitate a person’s voiceincluding your emotion and intonation, after only three seconds of training.
This technology has been called VALLEY and we have known about it thanks to a 15-page writing published a few months ago in arXiv.
Those from Redmond were already using AI since before this project through nuancea company that they bought for 20 million dollars in 2022, with which they are also experimenting with voice and TTS.
In addition, they also use technology from startup OpenAI, including their ChatGPT tool, to implement it in their bing engine and its Office suite.
What all these projects and experiments conclude is that the perception of the human voice has improved a lot, since they no longer seem like robots like in the past, but there is still a way to go to resemble one hundred percent what human voices are. .
VALL-E is a technology that unfortunately we cannot test, but we do have a good number of demos on GitHub.
ElevenLabs
For a few weeks we have been able to enter ElevenLabs and experiment with a system that demonstrates what progress has been made in technologies to clone the voice using Artificial Intelligence.
In this case, the company “brings out its chest” with respect to the processing of the audio clips that it is capable of generating.
Initially no restrictions to test the system, nor cuts for using it with inappropriate comments, but it is true that the company has declared that it has observed “a growing number of cases of misuse of voice cloning” so they reserve to take action.
The most important thing about ElevenLabs is that it can be tested, so it is extremely interesting.
It should be noted that, to work with this system, we must sign up for the first paid version (5 dollars a month) that offers a first month freeallowing us to unsubscribe before it expires so as not to pay anything if the system has not convinced us.
To do this we will do the following:
- The first thing is to register on the ElevenLabs website, clicking on sign up. We can do it through a e-mail or using Googlein addition to Facebook.
- Then we will see how they exist different plansfrom a free one, in which voice cloning does not work, through the Startet for 5 euros per month, up to one for businesses that exceeds 300 dollars per month.
- As we already told you before, we must subscribe to the Started plan that has a free month to try voice cloning.
- Now we click on Voice Lab and in the dropdown we do it in Voice Cloning.
- It is time to start with our first voice by clicking on Add Instant Voice.
- We can get on the platform a Audio with a voice recording (must be longer than 1 minute).
- After loading the MP3 file with the voice we press add voiceputting beforehand a name for said voice in Yam.
- Now we will see how the voice appears at the bottom loaded.
- At that moment we click on Edit and it will take us to another window where we must write the text that we want the system to reproduce.
- We must warn you that, for now, it is optimized for the american english accentso it may be the case that you half recognize your voice.
- Obviously, if you are used to hearing yourself in English, you can try writing the text in Shakespeare’s language, otherwise you will hear yourself as if you had been born in the homeland of McDonald’s hamburgers.
It is a good way to have our voice cloned and able to say what we want.
You can tell us what you think of the two most cutting-edge systems in terms of voice cloning through AI on our social networks. We would like to know your opinion.