Describe the song: a mood, a place, a story or a style. Leave empty for a surprise.
Band: style lets a Yamaha arranger style play the backing band, band uses our own drums, bass and keys, auto lets the prompt decide.
Lyrics to sing, with optional [verse], [chorus] and [bridge] tags. The song takes its form from them. Leave empty to have them written.
What the written lyrics should be about. Leave empty to use the song prompt.
Vocals: tick Sing it and pick a Singer , or leave it on auto to follow the prompt. Upload a clip of clean singing or speech (10 to 20 seconds, a voice you have the right to use) to sing in that voice, or leave it empty for the default voice. male , female and female breathy are SoulX-Singer voices; the named singers are DiffSinger voicebanks. Language picks the lyrics' language; auto lets a named singer sing its native language. Native Japanese: Aka Subaru, Amaboshi Cipher, Denka Shaa, Gahata Meiji, Karasu Yuutsukoe, Kuro Bousuku, Monone, Natsume Juli; native English: Nishiren; the Russian and German singers sing English here. Japanese and Chinese need a DiffSinger singer, and SoulX-Singer sings English and Chinese.
MIDI use: similar writes a new song over its chords and form, cover keeps its chords and form and has an arranger rework its melody and band, clean plays its parts tightened to the beat with our sounds and mix, use plays its parts as they are.
Length of the song, 30 to 600 seconds (10 minutes).
Beats per minute, 60 to 180. 0 picks one that fits the genre.
0 to 100. Higher values stray further from the obvious choices.