Step-by-Step Guide to Creating Music with Descript

Ross Zeiger shows how to use Descript's AI music generator: write a text prompt with genre, mood, and lyrics, then generate audio in seconds. One credit costs roughly ten seconds of audio. He also covers sound effects, warns that AI-generated audio ownership is unclear and may get flagged, and recommends legal advice before commercial use.
Chapters
Descript just added the ability to generate music using AI and a text prompt to put music into your videos and customize that music in terms of the genre, the lyrics, the length, and more. So here's an example. This is a prompt that I gave it. I said, create a song to be played at the intro of a Descript tutorial. The topic of the tutorial is how to generate music for your videos using Descript. It should be funny. And I ran this multiple times to get multiple different results. So I'll just play a couple samples of each of these so you can start with the end in mind.
Need a beat for your video scene, but you don't even know what a synth should mean.
And then here's another one.
Staring at a video that has no sound? Let's turn this quiet track around. Just click on these click and jam.
So you can see every single time I run it, it gives a slightly different result and they're very cheap on the AI credits. I'll talk about that at the end. So as far as how to actually do this, we start by going to the AI tools button on the right panel. And under the Sound Good category, you'll see the Generate Audio button. Click on that. And at the very top is a prompt. And so this is where you're going to put in what you want it to make. And you can make the prompt, the box a little bit bigger by clicking the bottom right corner and dragging it down. And so you say, like I did here, create a song to be played at the intro of a Descript tutorial. The topic of the tutorial is how to generate music for your videos using Descript. It should be funny. So describe the genre you want. What is it? What kind of genre is it inspired by? Is it jazz? Is it classical, et cetera? And then a mood for the lyrics or for the pacing, et cetera. So you can type all that in here, or you can click this Inspire Me button and it's just going to put in some generic pre-made prompts like energetic pop instrumental with driving drums and bright synths. So I'm going to leave that in there. And then below that, there's a thing that says auto, and this is going to be for the pacing of the video. So slow is 60 to 90 beats per minute, moderate all the way up to very fast at 150 to 180 beats per minute. Or you can just leave it on auto and it'll decide for you. And then next to that, you can choose a duration. So 3 seconds is the shortest you can generate. 15 seconds is the next. 30 seconds, all the way up to 10 minutes. So based on where you're going to use this, choose accordingly. Now, as for the credits that this uses, I did the 30-second generations, and I can show you if I click on the credit monitor up here in the top right, I can see that the sound generation right there was 3 credits. So it's basically 1 credit per 10 seconds of generated music. And it generates extremely fast once you actually do it. So anyways, pretty light on the credits, I would say, and pretty efficient. Next, you can choose to include lyrics. So if you toggle that off, you'll just get an instrumental song where it'll just be the, the instrument only. Or if you toggle it on, you can actually paste in there the lyrics that you want. You can give it some ideas like make it funny or What I've been doing is just leaving it blank and letting the AI decide, come up with the lyrics for me. And then lastly, there is a more options box where you can type in a genre. So I'll type in hip-hop, for example. You can choose a mood and you can say avoid. So this would be things that you don't want it to do, like put in drums, put in heavy bass, whatever you want it to avoid doing. Doing. And then you just click Generate. And let's see how long this actually takes here. Okay, that took 8 seconds to generate a 30-second song. Pretty incredible. And then you can hit this play button. Feel the electricity, let it take control. Brightness rising up deep inside your soul. And that's the song it came up with. And then there's a little download button, so you could download this to your computer. Or if you want to go ahead and add it to your project, you click on the add to project button right there. And it drops it into your Descript project as a layer. So from there, you would just move this around to fit exactly where you want it to fit with the rest of your video. And you can go to properties. You can adjust the volume level. You can mute it if you want. You can have it loop. All the stuff that you can normally do. With an audio layer, you can do with this music layer that you've generated. If we return to our Generate Audio tab, the other thing you can do is— everything I just showed you was generating music. To the right of that, there's another tab, which is Generate Sound Effects. And over here you have the same sort of thing. You enter in a prompt. So I'm going to say airplane taking off. And then you choose a length. I'm gonna say 5 seconds. This one is if you want the audio to loop or not. I'm just gonna turn that off and I'm gonna hit generate. And here it is. That's pretty good. And so once again, you could hit that little plus button to add it as a layer to your project, or you could hit that little download button to download it to your computer. And voilà, you have your own bespoke sound effect. Now, something that's important to mention is the copyright considerations of doing this. As you probably know, everything in AI is trained off of a corpus of existing human-made data, and generating music and sound effects is no different. Therefore, there are copyright considerations, and it's not black and white, and this is not legal advice. So my general advice here is just be careful of how you're using it. If you're in, if you're at all in doubt, if you're making money off of the content, probably don't use it. But they have here on their help article that Descript doesn't claim ownership of the music or sound effects you generate in Descript. However, the copyright treatment of AI-generated audio is uncertain. All this stuff is brand new and it's still being figured out. In the courts. So some or all of that music or these sound effects may not be eligible for copyright protection. Copyright rules can vary by country and outcomes can change depending on the specific facts and circumstances surrounding the creation of the work. Consult a legal advisor if you need guidance for a specific project. And it also says here, if you upload your finished content to another platform, that platform may still run its own copyright or policy checks. So if you're using this music on YouTube or Instagram or whatever, you still might even flag a copyright restriction. So be aware of that. This isn't completely safe from copyright infringement. The other thing Descript's AI can help you with is generating your thumbnails for YouTube. If you want to learn how to do that, click on this link right here on your screen.










