Table of Contents
If you’ve been keeping up with recent news, you might have noticed a growing trend among fans: using AI technology to transform their own voices or the voices of famous artists into something truly extraordinary. Imagine Drake, The Weeknd, or even Liam Gallagher from Oasis singing your favorite songs. It sounds fascinating, right? This advancement in AI has significant implications for the music industry. While only a handful of examples have caught the public’s attention so far, thousands of fans have already used AI to reimagine popular songs with new voices.
The Magic Behind AI Cover Songs
You might be wondering how this AI cover song magic works. The most popular method involves using open-source software called SoftVC VITS Singing Voice Conversion, or So-VITS-SVC for short. This AI-powered deep learning model can be trained using audio files of any vocal timbre. By utilizing vocal stems from an artist’s recordings, So-VITS-SVC learns to convert any vocal recording into the singing voice of that artist.
Accessing the Software
To access So-VITS-SVC, you have a couple of options. You can download and run the software locally, but this requires some coding knowledge and a powerful GPU. Alternatively, you can visit AI World, a popular Discord server for AI vocal experiments, and use their automated bot. AI World’s bot simplifies the process and makes it accessible to everyone. (Please note that as of 07/12, AI World’s bot is temporarily down for maintenance, but it will be back online soon.)
Before you begin, make sure you have a clean a cappella vocal recording. There are paid tools like RipX that can lift vocals from full mixes, but you can also try free options like VocalRemover. Once you have your a cappella file in WAV, MP3, MOV, or MP4 format, head over to the “ai-bot-1” channel in the AI World server and drop your audio file into the chatbox.
Emulate Your Favorite Artist
In your message, accompany the file with the text “-model @JuiceAI1,” replacing “model” with the name of the voice you want to emulate. AI World currently offers 18 voices to choose from, including Kanye West, Drake, Ariana Grande, Taylor Swift, Justin Bieber, Kendrick Lamar, and Travis Scott, among others. You can find the exact model names in the “available-voices” channel in the Discord sidebar.
Once you’ve specified the model and attached the file, hit enter, and the bot will process your audio. After downloading the processed audio, you can lay it over the instrumental version of your chosen track in a DAW like Logic Pro or Ableton Live. Adjust the relative levels of each track for a balanced mix, and then bounce the final product.
Exploring More Voices and Options
If you want to experiment with voices beyond the selection available in the AI World Discord, you have a few options. You can download and install So-VITS-SVC and make use of the ever-growing number of voice models created by the online community. From Chris Cornell to Chief Keef, you’ll find a variety of names regularly updated.
Another way to access AI voice conversion without installing So-VITS-SVC is by taking advantage of community-created Google Colab notebooks. These notebooks will run the necessary software on Google’s cloud servers, eliminating the need for local installation. Follow the instructions and click the grey “play” buttons to advance through the process.
Paid Alternatives
If you’re looking for a more streamlined experience, several paid services offer AI vocal conversion. Uberduck can imitate voices ranging from David Bowie, Brian Wilson, and Marvin Gaye to Lady Gaga, Kim Petras, and Grimes. You can even create your own models with custom datasets. While you can test Uberduck for free, voice cloning requires a subscription.
Voicify is another alternative that offers AI covers using models trained on artists like Drake, Juice WRLD, Pop Smoke, and Michael Jackson. Although the range of available voices is smaller compared to Uberduck, Voicify’s subscription plan is more affordable, ranging from $8.99 to $89.99 per month. Furthermore, Voicify allows you to create new models using any voice, as long as you have the necessary training data.
Legal Considerations
Before we wrap up, it’s essential to mention the legal status of AI covers, which is still unclear. An AI-generated song by Ghostwriter977, featuring Drake and The Weeknd, gained millions of views and streams soon after being uploaded. However, it was later taken down due to a copyright claim. So, we recommend exercising caution. While it’s not advisable to monetize AI-generated music that imitates existing artists’ voices on platforms like Spotify or YouTube, there’s no harm in having some fun with this technology at home.
If you want to learn more about AI vocal processing, the best places to start are the Discord servers AI World and AI Hub.
Don’t miss out on the latest gear news, artist interviews, and tech tutorials by following MusicRadar on Instagram!
