theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-04-24 06:46:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

acestep t2m

Browse Source
This commit is contained in:
mi804
2026-04-21 13:16:15 +08:00
parent a604d76339
commit 9d09e0431c
9 changed files with 300 additions and 377 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
diffsynth/models/ace_step_vae.py
View File

@@ -226,6 +226,7 @@ class AceStepVAE(nn.Module):
upsampling_ratios=upsampling_ratios,
channel_multiples=channel_multiples,
)
self.sampling_rate = sampling_rate
def encode(self, x: torch.Tensor) -> torch.Tensor:
"""Audio waveform [B, audio_channels, T] → latent [B, encoder_hidden_size, T']."""