Faisal87/radio-imaging-audio-generator

📻 Radio Imaging Audio Generator

Make radio jingles, station IDs and sweepers from a text description, with Meta's MusicGen running on a free GPU (ZeroGPU).

• —No OpenAI key needed: write the description yourself, or use the prompt builder.
• —Up to 3 takes per click, 3 to 30 seconds each, with a seed to make a take again.
• —Broadcast-ready loudness: EBU R128 (-23 LUFS) or online (-16 LUFS), with a look-ahead peak limiter so no sample goes above -1 dBFS.
• —WAV and MP3 downloads.
• —Optional: let GPT write a richer description, with your own OpenAI key (used once, never stored).

Based on Radio Imaging Audio Generator by Bilel Aroua (Bilsimaging), MIT License. This version by Muhammad Faisal. Source code, tests and measurements: GitHub.