Skip to content

Pretrained model for MIDI-less mode
Compare
Choose a tag to compare
@yqzhishen yqzhishen released this 25 Dec 13:03
· 215 commits to refactor since this release
v1.4.0
3d758c1

1215_opencpop_ds1000_fix_label_nomidi

MIDI-less mode, strict pinyin dictionary, 44.1kHz sampling rate, fixed some phoneme label errors and trained for 320k steps.
Note: both ph_dur and f0_seq should be given to run inference.

Assets 3
Loading