ltx-audio: use radix-2 fft to speed up compute_log_mel_spectrogram#1514
Open
stduhpf wants to merge 1 commit into
Open
ltx-audio: use radix-2 fft to speed up compute_log_mel_spectrogram#1514stduhpf wants to merge 1 commit into
stduhpf wants to merge 1 commit into
Conversation
Contributor
Author
|
Even with this change, the audio decoding remains very slow, especially for longer sequences. Maybe it would be worth investigating for using temporal tiling there too? |
Owner
|
I plan to add the missing operator to ggml, which should make it much faster. If this work takes too long or there are issues with the implementation, I’ll consider merging this PR. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR optimises one part of the LTX2.3 audio VAE by implementing Cooley-Tukey Radix-2 FFT algorithm to replace inlined DFT. Output sound quality is indistinguishable in my experience (as it should be).
(I also changed the intermediate
mel_valueto be float instead of double, this technically increases the footprint of this PR a bit, I don't think it affect performance or quality much either way, I kinda included it in this PR by mistake, but the code is a bit cleaner like that, so I'm leaving it in)Related Issue / Discussion
Additional Information
Performace comparison, generating a 32x32 24fps video
Checklist