[M1] Real Audio Readiness

[kymuco]

Goal

Make PracticeLens honest and useful on real musician recordings, not only generated demo data.

This milestone answers the product question:

What happens when a musician gives PracticeLens real guitar or vocal recordings?

Scope

M1 should improve documentation, suitability checks, diagnostics, and confidence warnings for real recordings.

It should not add heavy ML/DL, polyphony, tutor mode, cloud infrastructure, or HDE integration yet.

PR sequence

• [M1][PR1.1] Real audio usage guide — PR1.1: add real audio usage guide #88
• [M1][PR1.2] Input suitability summary — PR1.2: add input suitability summary #91
• [M1][PR1.3] Duration mismatch diagnostic — PR1.3: add duration mismatch diagnostic #92
• [M1][PR1.4] Leading silence / start offset diagnostic — PR1.4: add start-region input suitability diagnostic #93
• [M1][PR1.5] Low-confidence warnings in Markdown — PR1.5: render low-confidence warnings in Markdown #94
• [M1][PR1.6] Real audio smoke workflow docs — PR1.6: document real audio smoke workflow #95
• [M1][PR1.7] Real-audio manual checklist — PR1.7: add real-audio manual trust checklist #96

Definition of Done

• A real user can understand how to record reference/takes.
• Reports warn when analysis confidence is weak.
• Obvious real-recording issues are surfaced instead of silently producing overconfident feedback.
• No new broad product surface is introduced before the core real-audio loop is trustworthy.

Result

M1 is complete.

Implemented through merged PRs #88, #91, #92, #93, #94, #95, and #96.

[kymuco]

Closing this tracking issue because M1 is now represented by a real GitHub Milestone.