Replace lyrics in MIDI JSON with new text and auto-generate phonemes. Smart matching: splits lyrics by newlines/punctuation into sentences, each mapped to a MIDI section. If new sentence is shorter than original, extra characters keep original text. If longer, automatically splits the longest-duration note (pitch unchanged, duration halved) to create more positions. Consecutive repeated characters in original are grouped (e.g. '向向' → slot with count 2) and expanded accordingly. Chinese chars → zh_ pinyin; English → en_ phonemes. Force Tone 4: forces Chinese phonemes at pitch >= threshold to tone 4. Fixed Pause (default ON): keep SP durations unchanged. Flexible Pause: when tokens are crowded or SP is overly long, redistribute SP time proportionally to tokens. Split Mode: 'token' = allocate chars by original token count proportion; 'duration' = allocate chars by original section duration proportion. Speed: adjust playback speed (1.0 = normal, >1 faster, <1 slower). Duration and f0 are scaled proportionally.
By ahkimkoo·Created 2 months ago·Updated 2 months ago· 3