The extra note stacked on the fifth eighth was consuming the next sequence step, so later bars drifted. Drop that trigger instead of delaying the rest of the cycle.
Co-authored-by: Cursor <cursoragent@cursor.com>
The circled Em notes were stacked on the previous eighth. Hold only that extra note, and time it from the millisecond clock that actually runs.
Co-authored-by: Cursor <cursoragent@cursor.com>
III-VII shapes use the C-major sequence column with the same fixed-key offset as the strings. Chord tones follow the decimal column. Bump firmware to 0.2.57.
Co-authored-by: Cursor <cursoragent@cursor.com>
Bump firmware to 0.2.44 so slot 1 stays Bank1 program 24 and slot 2 stays program 25. Rhythm presets no longer replace the string voice on channel 0.
Co-authored-by: Cursor <cursoragent@cursor.com>
G-major III was collapsing B3 onto B2 after register -12 plus transpose -12.
Keep single register drop for III/IV; stack xpose only from V upward.
Co-authored-by: Cursor <cursoragent@cursor.com>
Match 转调说明 table: C~F keep normal register; #F~B lower bass/chord by one octave for all fret degrees. III+ register fix unchanged. RTT pitch test covers I/II+#F.
Co-authored-by: Cursor <cursoragent@cursor.com>
Match dual-bank tone ROM so accompaniment volume scaling no longer shares patches with the main string voice; bass program updated to 34.
Co-authored-by: Cursor <cursoragent@cursor.com>
TMR7 high-prio watchdog and fault-loop emergency cut/reset so soft-off still works if the main loop or AutoBand ISR dies; skip start when preset load fails.
Co-authored-by: Cursor <cursoragent@cursor.com>
AutoBand ignores input octave, so shift engine MIDI output by -12 for those chords (except drums) to keep the range from low #IV up to natural IV.
Co-authored-by: Cursor <cursoragent@cursor.com>
Log Flash magic on load failure, expose RTT flash bin3, and document the 0903 pack offset for the universal bank.
Co-authored-by: Cursor <cursoragent@cursor.com>
yuquanjun