I’m building ScanBass, a small MVP that lets a user drag-and-drop an audio clip of up to 30 seconds, automatically extracts the bassline, and returns a notes-only MIDI file. I don’t need a brand-new model; combining proven open-source blocks such as Demucs, Basic Pitch, or any comparable tool is fine. What matters most is a user-friendly interface. A simple, responsive frontend in plain HTML with Tailwind or a Gradio UI meets the need. On the back end, FastAPI is preferred, yet keeping the whole flow in Gradio is acceptable if it simplifies delivery. Core scope • Accept WAV or MP3 upload and show clear progress while processing. • Separate stems, detect the bassline, and convert it to MIDI. • Offer the finished .mid file for download. • Provide clean, readable code plus a straight-forward README for local setup (venv or Docker). • Optional but welcome: deploy the working prototype to my Render.com account and supply the brief steps you followed. Deliverables 1. Source code and environment files. 2. Running prototype demonstrating the full flow end to end. 3. Short technical note covering key decisions, dependencies, and ideas for improvement. When you reply, please outline: • How long you expect to deliver this functional prototype. • Your price proposal that fits my 15–25 USD per hour expectation. • Any relevant experience with Demucs, Basic Pitch, audio ML, FastAPI, or Gradio. I’m ready to move forward quickly and will review offers as they arrive.