30s clip
1 credit
30 seconds. MP3 mix only, plus a license file.
Make ad music with AI
Describe the product, the feeling the ad should leave, and the edit length. Use Instrumental under a voiceover. Test directions with 30-second clips, then render a studio WAV master for the final mix. Commercial use follows each song's license file.
Start from a preset
Pick one to load it into the composer. Choose the output type there, then edit the prompt to make it yours.
Clip, full song, or studio WAV?
Full song
10 credits
Auto, 1, 2, or 3 minutes in the composer (up to 4 via the API). MP3 mix, stems, and a streaming master, plus a license file.
Studio WAV
Recommended20 credits
Auto, 1, 2, or 3 minutes in the composer (up to 4 via the API). WAV 44.1kHz stereo mix, stems, and a streaming master, plus a license file.
Editors want a lossless file and stems to duck under voiceover. The studio WAV gives both. Use clips first to agree on a direction with the client.
Signup includes 10 free credits. You get stems and a streaming master on full songs and studio WAVs (not 30-second clips). See pricing.
Licensing
Ads are commercial use. That use follows the per-song license, and Lyria tracks are also subject to Google Gemini paid API terms. Read the licensing guide before the spot airs.
Read the commercial use and licensing guide for YouTube, ads, podcasts, and client work. The terms of service are the legal source.
Questions
Can I use the music in a paid campaign?
Commercial use follows the per-song license file. See the licensing guide for ads and client work.
Can I hit a specific edit length?
Pick 1, 2, or 3 minutes in the composer (the API accepts up to 4), then cut to length in your editor. Clips are a fixed 30 seconds.
How do I prove where the music came from?
Every song ships a license.txt with a signature bound to the song ID, engine, and date. The licensing guide explains how to verify it.
Are the stems fully separated?
Not all four. Vocals and drums are separated from the mix; bass is an 80 Hz amplitude-modulated reference derived from the vocal envelope, and other is a side-channel residual. Those two are derived reference tracks, not separated stems. The README in the stems zip explains each file.