Imagine being able to use the voice of Michael Jackson, Freddie Mercury, Beyoncé or Britney Spears on your own music, it seems unrealistic. However, it is possible by using an A.I.! Here’s how to do it:
To apply an artist’s voice to a song, you first need to create a template file using ‘Diff-SVC’ with an acapella file of the singer (Voice only without reverb or instruments). Then you will be able to convert another acapella file into the voice of the template. Online services like Uberduck or Voicify allow you to automate all these processes for a fee.
Warning: You must ask for permission from the people and authors before using their voices and creations. You alone are responsible for the consequences of your actions with the tools presented here! This article is purely informative and educational.
Which online tools or services should you use? Let’s take a look..
How to replace your voice for free with an A.I.?
It is quite possible to use A.I. voice modification tools for free and you won’t need to install or pay anything!
You will need to use the Google Colab platform with your Google account to install the SO-VITS-SVC tool in a few clicks. Run the tasks one by one by pressing the ‘Play’ buttons, wait for each script to finish before running the next one.
Then you can convert your voice (and vocals) after uploading the WAV file of your voice to Google Collab to use templates with the voices of Kanye West or Drake for example. You can download other user-designed templates on this Discord and use them on SO-VITS-SVC.
To find acapella voices to convert into a template, you can go to www.acapellas4u.co.uk or make your own acapella with the technique described in the following video.
If you’re not the type of person who likes to get technical, you can use an online service called Uberduck to convert your voice into a celebrity’s (for a subscription fee).
After signing up, you’ll get 300 free demo credits to try out the different voices available.
From Ariana Grande to The Weeknd, David Bowie or Lady Gaga, you will be able to transform your vocal line and apply the voice of one of your favorite stars.
You’ll need to make a clean recording without reverb, with perhaps a little compression and normalize the volume to -1dB so that the algorithm can work properly.
If you are a rapper, you can use the tool to replace your voice with Kanye West or Eminem’s for example, without having to sing. You can even do a Rap-Battle between Joe Biden and Donald Trump for fun 😁.
It will also be possible for you to do without voice recording beforehand, using Text-To-Speech, nevertheless you won’t have the possibility to make the voice sing, you will only be able to make it read text, but nothing prevents you from using it to generate samples to use in vox songs.
Here is a good example:
Another online service will allow you to convert your recordings to change the voice, it is called Voicify. An advantage over Uberduck is that it will allow you to convert the voice of a video by directly entering the link of the YouTube video, which is quite convenient.
An example with Kanye West singing “Wake Me Up” by Avicii:
As the service is also paid, it is better to choose your audio files carefully to convert or pay attention to the text you want to read, because the credits can go away quite quickly and mistakes are easy to make.
The case of Drake and The Weeknd
Unusual fact: A track generated by artificial intelligence named ‘Heart on My Sleeve‘ has aroused the interest of Internet users on TikTok and other social networks, which has displeased the record company of Drake and The Weeknd. The track has been removed on TikTok, but it is still possible to find it by searching a little on YouTube :
The controversy being that it is a synthesized voice of Drake and The Weeknd (created by an A.I.) but the track does not belong to the record company, so they should not be able to have the track removed from the platform.
The press has seized the subject, and evokes the legal hole in this case, since there is no law prohibiting or allowing such a use of artificial intelligence. However, it should be noted that the two artists did not agree to the reproduction of their voices, which could lead to legal proceedings against the author of the title for identity theft.
Even if the results are not perfect, A.I. still manages to make the dead sing, which will delight some and disappoint others. One thing is sure, this is only the beginning and the advances are so fast that it is possible that it will influence and greatly disrupt the future musical universe before the music industry has time to adapt or the governments vote laws governing the use of A.I.
We must not forget that behind a voice is a human being with his own sensitivity and artistic choices. To forget this would be to consider that his voice is only an ersatz of what he is or has been.