How to generate AI music with Stable Audio: the complete guide

What is Stable Audio?

Generators assisted by artificial intelligence will become widely available during the year 2023. From text-generating AIs, like ChatGPT, to those that create images, like Midjourney or DALL-E, including video generation (Gen-2), all areas of creation are affected, and audio is no exception. Several tools exist for generating music using AI, and Stable Audio is one of them.

Developed by Stability AI, already editor of Stable Diffusion for image generation, Stable Audio allows you to generate musical extracts – from orchestral sequences to instrument samples – but also sound effects. The paid versions of the tool also give you the rights to commercial use of the extracts you have created.

How to generate music with Stable Audio

1. Connect to the Stable Audio platform

Stable Audio allows non-paying users to generate up to 20 clips for free. To do this, you will need to create an account on the platform. Meet at this addressthen click at the top right of your screen on the button Try it out for free to create your account, or directly on Log in to connect you.

2. Generate music with Stable Audio

Once connected to Stable Audio, you can now generate music or sound effects. To do this, the process is similar to any AI generator. All you need to do is write a prompt in the dialog box provided for this purpose. Complete your request by simply asking Stable Audio to generate a theme music or sound effect for you, such as a singing birds tune or a drum solo, then click the arrow below the prompt window. In a few seconds, the tool will create your audio sequence.

What possibilities does Stable Audio offer?

Three main types of content

With Stable Audio, it is possible to generate three distinct types of content, explains the platform. You can create:

• Complete instrumental versions: use Stable Audio to generate an audio sequence including a range of instruments. The platform gives several examples of prompts for this, such as “ Lofi hip hop beat » Or “ Calm meditation music to play in a spa lobby »and encourages you to include as much detail as possible.
• Stems or samples of instrument(s): you can generate a stem or sample from a single instrument or group of instruments, such as “ Samba percussion », “Drum solo” or “ Electric guitar top line instrumental solo, no drums, classic rock, 105 BPM ».
• Sound effects: Stable Audio can also be used to create sound effects, such as a phone ringing, the sound of an explosion, a car passing on the street, birds singing, etc.

How to write your prompt well

For the moment on Stable Audio, the writing of a prompt is done in English only, even if many musical terms are close to our French. You will therefore first need to make a little translation effort to adapt your request. The platform sets up a little guide to help you write your prompts in the best possible way. “This is what works for us, explains Stable Audio, and we encourage you to experiment with the tool and find what works for you! » Here is the procedure proposed by Stable Audio:

1. Add details: musical genres, descriptive phrases, instruments and moods “work particularly well”.
2. Set the mood: to do this, you can combine musical and emotional terms. For example, an audio sequence can be “groovy” or “rhythmic”, and the emotion released can be “sad”, “nostalgic”, “celestial”.
3. Choosing instruments: Stable Audio finds that it is useful to add adjectives to instrument names, such as “powerful choir”, “vibrant strings”, etc.
4. Set the tempo: adjusting the BPM (beats per minute) will allow Stable Audio to properly define the tempo of your musical extract. The key is to try to stick to BPMs that suit the style of music you’re generating. For example, for an instrumental hip hop extract in a “boom bap” style, tell the AI ​​that you want a BPM of 90.

Set up the final details

Now that you know how to write your prompt, all you have to do is handle the final details before generating the desired snippet and then uploading it. Several points should be noted:

• The subscription-free version of Stable Audio allows you to generate 20 clips for free, with a maximum duration of 45 seconds,
• Under the dialog box where you write your prompt, you can adjust the duration of your excerpt,
• Once your extract has been generated, you can download it in MP3 for free, or in WAV if you take advantage of a paid offer, by clicking on the button provided for this purpose under the generated extract or under the dialog box, in your library Creation.

Paid plans allow, for the $11.99/month offer, the generation of 500 extracts of a maximum duration of 90 seconds per month, and the downloading of your creation in WAV format; for the superior offer (price on quote), a personalized number of generations and maximum duration. These two paid offers offer commercial use of the extracts created.

Some examples of musical extracts created with Stable Audio

We experimented with the tool with several example prompts. All snippets were generated and uploaded with the free tier. Here are the results.

Prompt:Hip hop boom bap beat, MPC 1000, 90 BPM, east coast new york, beat, snares, charley, piano melody

Prompt: Post-Rock, Guitars, Drum Kit, Bass, Strings, Euphoric, Up-Lifting, Moody, Flowing, Raw, Epic, Sentimental, 125 BPM

Prompt: Calm western music, cinematic, soundtrack

Prompt: Crowd in stadium cheering