Suno is a world-renowned music generation model that can create very beautiful songs. Ordinary people can't tell whether a song was generated by Suno or actually composed by a human.
In simple terms, Suno has a cliff-edge leading model performance in the field of music generation. I believe that by reading this article, you will more or less understand some of Suno's features. In this article, I will mainly introduce how to use API pass to access Suno. Then use Suno's song generation capabilities.
Generate a song
First of all, unlike other APIs, Sonar has its own business scenarios. So, in order to better understand the APIs provided by APIpass, we first need to understand the existing scenarios of Suno.
We can access the official Create page of Suno, and you will find that it has four corresponding modes. This is also the most mainstream way for us to use Suno, which is the four different options for creating songs. In the image below, I have already marked with red arrows how these four modes are set up and activated on the official Suno page.
First, we have our simple mode. In simple mode, you can choose to enable or disable the instrumental. Just as its name implies, if you enable the instrumental mode, it will only be used to generate instrumental performances. To put it more bluntly, it is used to generate pure music, meaning there will be no vocals, no feeling of a singer singing, just pure music. This kind of scenario is usually suitable for when I want to relax, when I want to listen to some piano pieces, when I want to listen to some very calm, or say, rhythmically strong background music. Then we can use this mode.
We are referring to the documentation provided by API pass, specifically the page: https://apipass.dev/model/suno/suno_generate
We mainly look at its input.customMode parameter. In fact, if the custom mode parameter is false, it is actually the simple mode. So if it's true, then it's Suno's current Advanced mode.
So, back to our previous discussion, in simple mode, and with instrumental turned off, we need to set customer to false and instrumental to false when passing parameters. At this point, the song description on the page is equivalent to us only needing to pass the prompt parameter.
This is situation one. In situation two, it is still in the new mode, but the user has turned on instrumental and wants to generate pure music. At this point, you might notice that the layout of the page hasn't changed at all. On the Sonar homepage, only the song description parameter is passed to you. So, after we pass the meal, in Apipass, our custom mode remains false, but our instrumental will become true. Similarly, we only need to pass ApiPass. Here, the Apipass will affect the corresponding song description area on the page.
Similar to the simple mode, I turned off the simple mode and switched to the Advanced mode. In this mode, you can also choose whether to generate instrumental music or vocal singing music.
While we were writing this article, we discovered that the structure of the official Sonar page has changed. In the past, when users were in Advanced mode, there was also the Instrumental option. That is, users could explicitly choose whether to generate instrumental music, but now the page has hidden the instrumental option.
On the surface, it seems that Suno no longer supports generating instrumental music under the Adventure mode. But in reality, after testing, as long as we don't fill in the lyrics and only specify the style of the song, when we click generate, it produces instrumental music just like before. It's just that Suno might have changed its UI design; perhaps the Suno team thinks that hiding the instrumental option is a better UI/UX design. But personally, I don't think it's that good because by hiding an option, you require users to have what you might call implicit knowledge when using your product. They need to know that they have to leave it blank to generate the instrumental version, which I think isn't very intuitive. In short, I think this is not a very good design for page improvement.
To get back on track, although the page has changed and differs from our initial understanding when writing the article, the functionality corresponding to the page remains completely consistent. Therefore, our API pass parameters actually do not need to change at all; using them as usual will still be compatible with the effects brought by this page change.
Similarly, if you enable Advanced, then you need to pass true for the custom model in the API pass. Then, if you think you need to generate an instrumental song, you should enable the instrumental option, and you don't need to provide a prompt. So here, the prompt actually corresponds to the lyrics on the page. If you want to generate an instrumental on the Suno page, let me emphasize again, based on our own practice, you don't need to provide the lyrics. Just provide the style and title, and it will be instrumental music.
In summary, during the phase of searching and generating music, our parameters actually correspond to its four different states. So normally, we pass parameters according to the method in the document, and we can obtain the corresponding music.
And one more piece of practical advice is that when we actually call our API to generate music, I recommend using the callback method. Because it takes a relatively long time to generate a song, the input and interaction will also return the status of the generation process.
What is the state of generation? When I generate a song, it's the streaming file of my song's audio file. I will return it during the generation. Assuming a song takes 5 minutes to generate, it might return the first callback at around 30 seconds. This callback would contain the real-time audio file of the song being generated.
At this point, you just put this audio file into your product, and the user experience will be very good; they will think you have already completed the song. Although strictly speaking, it hasn't actually completed generating a full song. Assuming a song is 3 minutes long, it might take 5 minutes to generate. At this point, you might be able to let the user listen to it for just 30 seconds. The music he listens to might just be the first five or six seconds of the song. He listens while it generates. So when he finishes listening, it might be done generating. It's basically this kind of interaction format that will be more beneficial for the conversion of your product.