Skip to content

[bugfix] Fix STFT dtype mismatch #1419

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
SolitaryThinker merged 1 commit into from
Jun 1, 2026
Merged

[bugfix] Fix STFT dtype mismatch #1419

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion fastvideo/models/audio/ltx2_audio_vae.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1634,7 +1634,7 @@ def forward(self, y: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
y = y.unsqueeze(1)
left_pad = max(0, self.win_length - self.hop_length)
y = F.pad(y, (left_pad, 0))
spec = F.conv1d(y, self.forward_basis, stride=self.hop_length, padding=0)
spec = F.conv1d(y, self.forward_basis.to(y.dtype), stride=self.hop_length, padding=0)

@gemini-code-assist gemini-code-assist Bot Jun 1, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

Using self.forward_basis.to(y) is more robust than self.forward_basis.to(y.dtype) because it automatically matches both the device and the data type of y. This prevents potential runtime errors due to device mismatches if y and self.forward_basis happen to be on different devices.

Suggested change
spec = F.conv1d(y, self.forward_basis.to(y.dtype), stride=self.hop_length, padding=0)
spec = F.conv1d(y, self.forward_basis.to(y), stride=self.hop_length, padding=0)

n_freqs = spec.shape[1] // 2
real, imag = spec[:, :n_freqs], spec[:, n_freqs:]
magnitude = torch.sqrt(real**2 + imag**2)
Expand Down
Loading