What's new at AI Music API

When a generation is rejected or fails, you now receive the real reason instead of a generic error — and temporary failures are retried automatically.

More reliable uploads for cover and extend requests, with up-front validation and clearer failure messages.

Resolved a rare double-charge when upgrading a subscription. Plan changes now happen in place, keeping your credits intact.

Added Google Lyria 3 Pro for high-fidelity instrumental generation, with a dedicated API and landing page.

Introduced public pay-as-you-go credit packs and a higher-volume Scale plan, so you can match spend to your workload.

Audio-to-MIDI conversion (beta): turn a finished track into editable MIDI for your DAW.

Hardened account security — API keys are now hashed at rest, with first-class key revocation and expiry.

V5 and V5.5 music generation are now available, with accurate model labelling on every track.

Producer AI gained lyric-generation and music-video clip endpoints.

Automatic credit refunds whenever a generation times out or fails — you are never charged for a song you didn't receive.

Stem separation now returns partial results after a short grace period instead of failing the whole job.

Added VAT / Tax ID collection at checkout and downloadable invoices and receipts.

Webhook callbacks: get notified the moment a task finishes, so you no longer have to poll for results.

Added an asynchronous voice-creation endpoint for building custom voices.

A major platform upgrade brought faster page loads and a more stable foundation across the product.

Task completion times are now tracked and exposed, making it easier to monitor performance.

Faster, lighter task polling — lower latency and cost for high-frequency workloads.

Reworked uploads to eliminate dropped-connection failures on large files.

Producer API added a preview model and per-account custom credit consumption.

Custom plans for high-volume customers, including the option to lift concurrency limits.

Producer AI API documentation and landing pages are now available in multiple languages.

Richer structured data across the site for better search visibility, plus an llms.txt so AI assistants can understand the API.

Launched a brand-new homepage and product experience.

Expanded the editing toolkit: cover songs, extend tracks, replace a section, swap instruments, swap vocals, and create reusable voice personas.