🤯 As a product designer for 5+ years, and music composer/producer for 20+ years, in my personal opinion, this is one of the most important breakthroughs in AI-Generated music. I've have tested and used hundreds of software products and plugins but Cassette AI has pushed the boundaries of instrumental generation. Like any text prompting generative technology, it will take a little while to find the right combinations of descriptors in your prompt to get a sound closer to your liking. BUT, once you find an impressive prompt, you can create variations from there. With the right prompt, the outputs are novel and they provide a great starting point to create some amazing music, sound packs and sound fx. If you are curious, creative, have a little bit of skill, and have a decent imagination, this tool should 100x your productivity at minimum.
The UI/UX is simple enough to figure out and their customer service and support is extremely responsive on their discord. They listen to their users and try to incorporate quick fixes or helpful features quickly.
Cassette AI has given me the "woah, I just got the chills" moment like summer/fall 2022 when I first used Midjourney and GPT 3 in the early stages of generative ai content. And like Midjourney or GPT 3 in fall of 2022, there will be some artifacts. Some sound like microphone feedback, or volume changes, weird rhythms and random audio fades. I like to imagine it being an AI system trying to figure out how to create the sound you want. I have empathy for that since that's kind of what it's like producing for an artist in a way. But it seems like they've considered this and give a generous amount of output generations for a reasonable price. The price point is perfect for the value for you can get out of it.
✅ 5 PRO TIP💲 :
1. Keep a spreadsheet or note of the prompts that generated an output that you really liked. (it will help you achieve that level of sound more frequently)
2. If you hear artifacts or sounds you don't like, use their stem generator to separate some of the sounds. (Sometimes isolating the drums, bass, and others might eliminate the artifact)
3. If you are passively doing something else and have the time, you can generate pieces up to 5 mins, though it may take that long to generate it, you can keep doing else your were doing while it generates and return to something new. (Sometimes it's really worth it)
4. Keep an eye out for the suggested modifiers below your output. They can sometimes give you amazing ideas or introduce you to new descriptors that you haven't thought of to add to your next prompt.
5. Use "Refine" button to refine your prompt, but make sure that it conveys what you were trying to describe and feel free to add to that.
It's an exciting time in an exciting space right now and I'm hyped for what might be on the roadmap for the Cassette AI team!
Easily, this is my new favorite tool coming into 2024 and I use it everyday with no fail. 😎 Bravo! 👏🏿👏🏿👏🏿
Cheers! 🍻
CassetteAI
ViewCoder
Unpredictably.io