SoundGen (soundgen.ai) is an AI music studio built around one idea: a generated track should be the start of a session, not a finished file you can't change.
Type a prompt and SoundGen generates a full track, a loop or a one-shot sound effect, with lyrics or instrumental, and optional tempo and key targets. The result lands on a real multi-track timeline: audio, MIDI, bus, folder and VCA tracks, automation lanes, a session (clip-launch) view and a piano roll with 16 built-in instruments.
From there you can split any audio, generated or uploaded, into stems (4, or 6 in Deep mode), convert a melody or bass line to MIDI, detect key, tempo and chords, and mix with 64 built-in effects, sidechain routing, latency compensation and a mastering chain with a limiter and metering. Bounce up to 96 kHz / 32-bit float.
An AI agent works inside the session with 200+ tools across the timeline, mixer, MIDI and arrangement. It shows a plan before it acts, and every action can be undone.
SoundGen runs in the browser, with desktop apps for macOS (signed and notarized) and Windows. Developers get a REST API and an MCP server, so assistants such as Claude Desktop or Cursor can drive projects directly.
Pricing is pay-as-you-go: 100 free credits on signup with no card, then one-time credit packs that never expire. Generated audio comes with a royalty-free licence for personal and commercial use on free and paid credits alike (see Terms §4).
Free browser tools, no sign-up, at soundgen.ai/tools.
