New AI plugin TextToSample lets you generate samples for free
Samplab states that the tool’s goal is “certainly not to replace human musicians, which it’s not capable of doing anyway.”
Image: Samplab via YouTube
The developers behind AI-based sample editing software Samplab have released a new plugin that allows you to generate samples from texts — for free.
As its name suggests, TextToSample works by producing samples in response to the user’s text prompt.
All you have to do is describe the kind of sound you want — e.g. ‘trap with drum fill’, or ‘rock guitar solo’ — and the software will generate a sample based on your commands. Users can also drag and drop pre-existing audio files into the plugin along with a prompt and have it generate new samples.
According to Samplab, TextToSample uses Meta’s open-source, AI-powered text-to-music language model, MusicGen, and was trained using data provided by the algorithm.
“Our goal with TextToSample is to enable the producer community to experiment with state-of-the-art generative AI models,” reads the FAQ on Samplab’s website. “We want to show what’s currently possible with these models, especially when running them locally on your computer.”
The tool has a fairly simple UI which makes it extremely beginner-friendly, though it also comes with some interesting features, such as the “context” dial, which “allows you to adjust how many seconds of audio the model looks at when generating the next bit” (via MusicRadar).
Check out a demo of TextToSample below.
Samplab is but one of many companies that have jumped on the AI music-making software bandwagon in recent months. This May, Google announced the arrival of MusicLM, the company’s very own AI-powered tool that turns written text prompts into high-fidelity music.
Simply type in a prompt like “soulful jazz for a dinner party” or “a calming violin melody backed by a distorted guitar riff” and MusicLM will create two versions of the song for you. Users can then vote on which one of the two tracks they like better, which serves to “help improve the model”.
The tool is also capable of transforming whistled and hummed melodies to the style you want.
