Wait, lyrics from a recipe?! When "Kung Pao Chicken" meets musical theater, it instantly becomes a sensation. This song has quickly climbed into the top ten of the global AI music charts. Over the past few days, Suno, dubbed the musical version of ChatGPT, has taken the internet by storm.
△ Photo credit to netizen @vigorxu
Check out this vibe.
Also trending are the fresh take on "Shuidiao Getou" (When Will the Moon Be Clear and Bright) and the Kpop version of "The High-Speed Kpop Invasion into China"...
On music charts, you can see a variety of music genres and languages represented, which has many internet users hooked.
However, some netizens have already shared tutorials on "how to make a complete song longer than 2 minutes with Suno".
Let's take a look at how Suno is taking the world by storm.
There are primarily three customizable options: Lyrics, Styles, and Themes.
After clicking on "Create," Suno will autonomously generate two demos for you. If you're satisfied, you can continue generating songs. If not, you have the option to remix, similar to the mode used in image generation. The lyrics, style, and even the choice of instrumental or vocal can be randomly generated, and you can opt to include a theme or not. It's a highly flexible creative process!
Apart from that, you can also remix from the hot list, which is like directly copying someone else's prompt to generate a song.
For example, here's a homemade remix of "Kung Pao Chicken." Ladies and gentlemen, listen up. Compared to the original version, which one's better?
"Kung Pao Chicken" (Remix) by Quantum Beats, 2 minutes
Furthermore, it also supports operating as a creator to make similar albums.
With this whole process, many netizens are exclaiming: The era where everyone creates songs has arrived!
When paired with other AI tools, there's a lot of creative potential to explore. For instance, combining Claude-3-Opus for lyrics and composition creates a clear division of labor. It's common practice now to merge Midjourney and Runway to produce music videos. There's even been instances of users generating 3D rap artists, pairing them with songs composed by Suno, resulting in some solid AI-driven rap. The company behind it all has only been around for two years.
The latest release, Suno V3, offers even higher quality music compared to previous versions, spanning a wider range of styles and genres. With just a few simple words, you can craft a complete 2-minute song in seconds, and it supports any language.
Before the public release, the new version was rolled out to Pro and Premier members for testing and improvement purposes.
The team has revealed that V4 is currently in development. Apart from this, there has been a lot of discussion about the team behind it. Suno, founded just two years ago, has a founding team consisting of machine learning experts Mikey Shulman, Freyberg, Georg Kucsko, and Martin Camacho, two of whom are also seasoned music enthusiasts.
Before founding Suno, they all worked together at Kensho Technologies, a company primarily focused on using AI to solve complex business problems. During their time there, the four of them jointly developed AI speech transcription technology used to transcribe earnings conference calls of publicly traded companies. It was during this development process that they saw the potential of AI audio generation.
According to Mikey Shulman, audio is a continuous signal. High-quality audio typically has a sampling rate of 44kHz or 48kHz, meaning "48000 tokens per second".
Currently, Suno has a team of 12 employees who are dedicated to enabling a billion people to have a deeper engagement with music. They also emphasize that they do not intend to replace artists.
