Shadow Aim now remembers your training progression and music library, and makes Juniper's music controls more responsive.
- Four-mode guided baseline, including illustrated reactive tracking. The main Practice button follows approved coaching, resumes practice, and restores original retest settings across restarts.
- Saved music folder and resumable scans. Juniper receives current scan progress and automatically chooses mood, genre, or artist queues. Next and natural playback preserve the selection; specific songs continue through related tracks.
- Confirmed song titles and volume changes, stale-control protection, and separate Luna/low control handling. Live synthetic requests applied in about 2.7 seconds; network timing varies.
- Immediate spoken round measurements while Sol/medium prepares the detailed review. New shots record center/edge placement and directional bias; older positions remain unknown.
- Deduplicated interruption handling clears buffered speech without dropping fresh replies while waiting for a final transcript. Live listening quality still needs broader validation.
- Explicit unscored sensitivity warmup and scoring countdown.
Both archives were built and exercised on native Windows and Linux CI, including first-time setup, restart, gameplay, and actual music-control integration. Automated voice checks use synthetic audio; they do not validate every microphone or spoken request. The detailed live review check took about 30 seconds, so detailed coaching is not instant.
Extract the archive and run Start Shadow Aim. First launch handles setup automatically. Close an older running copy before starting the update. ChatGPT subscription and GPT-Live/Juniper access are required for AI features; ordinary practice works without them.