Generative music AI has been much slower to develop than large language models for text, but the potential of Suno cannot be ignored. Text: Technode
Author|Huang Chentypesetting|Li Ruizi
This article is expected to take 7 minutes to read
Over the past year, AI music generation tools have advanced by leaps and bounds. Like Meta's MusicGen and Google's MusicFX, they can generate music clips, inspirations, and even entire tracks on demand.
Last year, tech company Suno launched a generative audio AI base model that allows users to generate two pieces of music with lyrics for about 30 seconds with a simple text input. The gameplay is similar to the previous Midjourney, just type the "/chirp" command into Suno's Discord channel, and add a hint to tell it the style of music and lyrics you want.
Founded by former Meta and TikTok employees, the company's mission is to create a business where "everyone can make music." They developed an AI platform that allows users to input text prompts to generate music clips that are up to two minutes long (V3 version). You can specify the genre, style, instrument, and lyrics of your music, and even enter your own lyrics. The V2 version of the platform has already produced some impressive results, although the resulting music still sounds a bit bland. However, the V3 version of Suno is more powerful.
In an interview with Rolling Stone, Mikey Shulman, co-founder of Suno, said the company is committed to democratizing music production and making it accessible to everyone, regardless of their instrumental skills or musical talent.
Ordinary people can also "play" with music
Tracks created with Suno, these works contain all the basic elements - vocals, melodies, chords, etc., and the lyrics and titles are all created by ChatGPT, and even the accompanying album art is AI-generated.
Brian Hiatt, a senior writer for Rolling Stone magazine, recently visited Suno's office in Cambridge, Massachusetts, and tried his hand at writing a song. He typed in the description "Mississippi Delta blues solo, about a sad AI", and the program generated the entire song in 15 seconds. Other music AI programs, such as Google's MusicLM, excel at creating futuristic dance music that sounds heavily synthetic, Hiatt said.
Siatt chose the Mississippi Delta Blues because of its human touch, and he wanted to give Suno a problem. "It's the most natural thing I can think of. I think some people think I'm exaggerating. But I'm really not exaggerating. I was really surprised, upset, and scared. I still get goosebumps every time I listen to this song because it's so weird. It shouldn't exist. He added, "Another reason I'm blown away by it is that it's so ahead of its time." ”
Over the past year, Hiat has been reporting on AI technology in music, and some experts have told him that it will take a long time to achieve an interface that can get the full song by simply typing in a description. But with Suno, he began to break prejudices.
Although the Suno V3 Alpha still has some issues with music keys, beats, and mixing, the company encourages users to test new versions and provide feedback so that they can continuously improve the product. Suno AI emphasizes that the V3 Alpha can be more creative than the V2 version in terms of song structure, and sometimes different prompts are needed to achieve the best results.
Right now, in the free version, you can experience all the features of Nano. You can write 10 songs per day for 50 credits, and if you exceed that limit, you can upgrade to the Pro Plan for $10 per month (500 songs). In addition, Suno clearly states on its official website that only paid users can use the generated songs for commercial purposes.
Some worries
Although Suno does not allow users to generate songs in a particular artist's style or use the voices of real artists, thus avoiding potential copyright issues. But Hiat argues that Suno could face legal action for its algorithmic training practices. OpenAI, the company behind ChatGPT, is currently facing multiple lawsuits and is reportedly using copyrighted data to train its technology. However, the precedent for copyright owners to take legal action against AI companies is still developing.
On the other hand, the free version of Suno is not yet available for commercial use, but it is still possible to make songs to share with friends. Their ultimate goal is to make a large number of music consumers music producers, just like providing a personalized streaming service, where all content is original. "They think it's fun and people will be willing to pay $10 a month or whatever it is to play it. I have to say that while I'm still thinking about this kind of ethical question, it's really fun to play. ”
So, what does this mean for human musicians? Siatt mentioned that a major producer and songwriter's main manager told him that Suno could disrupt the market for songs in commercials and TV shows, which is known as sync licensing (European and American countries have set up audio-visual synchronization rights, that is, the right to use music and video content simultaneously, and only this permission is required to be obtained when producing background music. There is no such right in the copyright law of the mainland, and the types of rights involved are reproduction rights and filming rights. ）。 For many independent artists, income from simultaneous transactions is a lifeline in a difficult industry, and the end of this revenue stream can be financially devastating.
Suno makes it easy for non-professional musicians to create music by providing an easy-to-use interface and powerful music generation capabilities. This trend towards the popularization of technology has helped to stimulate the creativity of more people and increase the diversity and number of musical works. Although Suno is able to create technically high-quality music, the value of a work of art often lies in its depth and emotional expression. Whether AI can fully understand the information and cultural context of the times and simulate the creativity and emotions of human artists is still a question to be discussed.
As music AI becomes more advanced, there's no guarantee that it will replace those elusive things that people like their favorite artists. Hiatt scoffed in response to a technologist's prediction that AI-generated artists as popular as Taylor Swift would be in the next 10 years. "Because people have such a deep connection with artists that audio products are just part of the work that artists do. And I think that if you can't relate it to a person, it won't have the same effect. ”
Even so, Suno's efforts should be seen, as one practitioner who studies audio algorithms said: "The general public's requirements for music are still too strict, and they only look at the results and not the process." The chain of music production is very long, and the lyrics, composition, arrangement, mixing, performance, recording, each piece is a little different, which is very different for the ears who are used to commercial music. And now, it's only been a year since MusicLM music generation came out of the circle. If we put aside the production and just look at the content, at some point, I would even wonder if the song is from AI (Suno) or human. Although this probability is not very high (at present), as long as there is, there is a possibility that traditional creations will be replaced. ”
This article is an original article by TechNode reporters, and may not be reprinted without authorization.
--------THAN--------
Interactive topics
What do you think of ChatGPT in the music industry?
Come and leave your opinion in the comment section!
*If you want to get industry information and share your experience with like-minded technology enthusiasts, then scan the code to add "Dynamic Points" to join the group chat!
Wonderful article is worth recommending!
One-click forwarding, poke and watch!