text | Lei Technology
Since the advent of chatgpt, generative AI has brought us many surprises, and AI's influence on different industries continues to spread. Only three months have passed in 2024, and a number of new large AI models have emerged: sora, which generates text videos, kimi, which supports 2 million words long text processing, , and suno, the rising star of Vincent Audio.
"chatgpt in the music industry", "terrible AI music that may sweep the world", "subverting the music industry" , these are the true comments given by countless netizens and media after experiencing suno.
suno is a music generation model launched by the AI start-up company suno. Regardless of whether the user has music theory knowledge or not, they only need to enter simple prompt words, such as music style, music genre, lyrics content, timbre, etc., and wait a few seconds to quickly generate lyrics with lyrics. and 2 minutes of music to the beat.
It is worth noting that suno has recently launched the v3 version, which can generate broadcast-quality music for the first time, and adds richer music styles and genre options, such as classical music, jazz, hiphop, electronic and other new trends wind. The official stated that the v4 version is already under development and plans to launch new features.
Suno’s first experience: One click to create a song, the effect is amazing
Seeing this, Lei Technology opened the suno official website out of curiosity. The first thing we saw was a lot of music generated by suno, mainly English songs. After listening to a few random songs, I feel that the melody is quite good, which makes me have higher expectations for suno's performance.
has registered an account and we open the creation page. The overall interface is similar to chatgpt, with keyword input boxes, music genres, model selection and other options.
Without further ado, let’s officially start the creative journey of “music novice”. In order to reflect the strength of suno, Lei Technology specially entered a keyword in Chinese that even it could not understand: "describing the love story of King Kong and Godzilla in a ballad".
After waiting for a few seconds, Lei Technology's first song "The Love Story of King Kong and Godzilla" was completed. Judging from the lyrics, suno accurately identified the two keywords King Kong and Godzilla, and extended the description of battle and other scenes based on their monster identities.
clicked to play the song, and the result shocked me. I didn’t believe it was a song created with the keywords I gave. At least doesn't sound as jerky as AI generation. The lyrics rhyme and even come with harmony and segmentation. As a "music novice", Lei Technology feels that this song meets the requirements.
Leitech then generated several songs of different genres and themes in succession. After the novelty of wore off, Leitech found that the songs of the same genre generated by suno felt the same. Although the lyrics and melody were different, my personal feeling was that they were the same. The saliva songs often heard on Douyin are very similar.
In the process of checking the information, Lei Technology discovered that suno actually has advanced gameplay. In the customization mode of the creation interface, users can customize intro (intro or prelude), verse (poetry part/verse), chorus (chorus part/chorus), bridge (bridge part), outro (outro), etc. Part of the lyrics, and through keyword techniques, let AI understand user expressions.
Because it involves actual music theory knowledge, it is quite difficult for friends who do not understand music to understand it. Lei Technology summarizes it as: "Style + Emotion + Instrument + Rhythm + Vocal". If you think it is too troublesome to think of lyrics, but you are interested in lyrics, you can combine it with chatgpt and let AI generate lyric text that meets your requirements.
If you want to refer to the rhythm of an existing song, you need to enter the bpm (tempo) and key (high pitch) of the song into the keywords.
After some experience, Lei Technology believes that suno’s performance is amazing. Whether it is one-click generation for novices or customized generation for professionals, can generate high-quality songs in a very short time. Especially the excellent works of custom generation and exploration pages show us the infinite possibilities of suno. At least for now, no one can determine the upper limit of the suno v3 version. The latest works that continue to emerge every day are the best proof.
Source: suno
ai music is powerful, but "human music" can never be replaced
Many people may be curious about what kind of company can create such a magical suno v3. At present, the suno team has only been established for two years and has only 12 team members. Some team members previously served in technology companies such as meta, tiktok and kensho technologies.
In fact, before suno appeared, there were several AI music generation tools on the market, including dream track, jammable and project music genai. The main reason why suno can spread virally on the Internet is that it simplifies the steps for ordinary people to create songs. Users can automatically complete vocals, lyrics, style, music scores and other content with simple guidance.
This instantly narrowed the gap between ordinary people and professional music creators. Just as midjourney-generated images caused turmoil in the design industry, suno has also aroused vigilance among some musicians. Although Suno's current creative level is far from reaching a stage that can subvert the music industry, the most terrifying thing about AI is its learning ability. The suno team only added the vocal music function to the generative model in July 2023. In just the past nine months, it has been iterated to the v3 version. Perhaps no one can predict the final level of suno.
According to Lei Technology, suno can indeed allow ordinary people to create "original" songs in large quantities, but the possibility of subverting the music industry is not high.
First of all, the essence of songs is people's self-expression, which is the same as words and pictures. However, the expression form of songs is more complex, and there are several possibilities just by the sound level. Of course,
ai can restore various sounds and styles through intensive learning, and even form coherent long music in the future.
But the reason why a good song resonates with the audience is because it is fully connected with people and society. This is also why we get excited and sad when listening to some songs. and AI music have not yet shown the corresponding capabilities. In view of the current situation, I prefer to call it a "ruthless creation machine".
Picture source: suno
Secondly, there is the commonplace copyright issue. The suno team has not announced what data the suno model is trained on. If they use copyrighted works without permission, they will face prosecution, and the music works generated by users with the help of suno also have the same risks of. In addition to generating new songs, many users of
also re-create existing songs. Whether this part of the operation is legal or not has not yet been determined. AI-generated content has brought many problems to the existing legal system. The birth of AI web texts, AI images, AI music, AI resurrection and other technologies has brought new complexities to intellectual property rights. How to avoid legal risks, be legal and ethical Guiding users to use suno to make music is the primary problem hindering the development of suno.
In fact, the suno team is also aware of the complex relationship between AI music and the music circle. They said that the team is committed to allowing people to have in-depth contact with music creation, rather than replacing musicians.
Finally, the real subversion of suno should be the production companies that specialize in creating online songs. Formulaic song creation has always been the strength of these companies, but suno clearly has an advantage over them. Under the impact of suno, this industry will usher in a new round of reshuffle.
suno is accidental, aigc's reconstruction of the content industry is inevitable. The
large model has not been in people's sight for a long time, but it has brought tangible changes to people's lives. Higher efficiency and lower threshold are the biggest charms of large models. In the "ai+x" scenario, the barriers to entry in the industry in the past are gone, everyone can be a creator, and everyone can express themselves to their heart's content.
suno, like chatgpt, sora, kimi and many other predecessors, has successfully caused a shock in the corresponding industry. Although still cannot meet the requirements for humans to output actual emotions, its song generation efficiency has successfully defeated 99% of musicians, and this advantage will continue to expand.
Lei Technology has not been using suno for a long time, but with the help of the guide, he has been able to create some songs that look a bit professional. Even if this is the case for "music novices", professional music creators will definitely use suno more efficiently and will naturally get more surprises.
When AI music becomes rampant, how users find music that suits their own aesthetics in the ocean of songs may become a new problem. The new generation of AI music recommendation that combines large models with music content will also usher in new opportunities.
suno and the aigc platform it represents are reshaping the order of the content industry.
Lei Technology thought of Douyin and tiktok. As can be seen from the name, Douyin attaches great importance to the value of "music" in short video content. A large part of tiktok's predecessor business originated from musical.ly acquired by Byte. It is no exaggeration to say that both Douyin and TikTok built short video empires based on "music", which is a characteristic that short video platforms such as Kuaishou do not have. Nowadays, Douyin has actually become an Internet celebrity music production machine.
Therefore, the emergence of suno and the outbreak of aigc should have the most direct impact on short video content platforms such as Douyin, because the production logic of content is undergoing drastic changes. Perhaps it is because of this that the former CEO of Douyin Group, which single-handedly made Douyin big, will resign and focus on film editing. The reserves and strength of bytes on AIGC cannot match its size, nor can it compete with giants such as Microsoft, Google, Meta, Baidu, and Alibaba. Fortunately for , Byte is increasing its focus on AIGC, because AI is Byte’s gene, and AIGC is the battle that Byte cannot lose. As for games, education, Feishu and other businesses, they are not part of the core. The wave of
aigc is coming at a speed that exceeds everyone’s expectations.
