VSE and Sound state / future

You do realize that there’s actual people working on this? Blender is not some nebulous ominous ‘them’. It’s just a group of people trying to decide on how to allocate the available manpower in the most beneficial way and not write themselves into a corner in the process.

Yes, the VSE could use some love. But it is quite a difficult beast to unify (ui-wise) with the rest of blender so it’s not strange it’s kinda hard to do it right. And there is someone actively working on the VSE right now.


I too wasn’t too happy with reverting the retiming tool, but honestly it was barely usable in state it was. It was much better than speed effect, but it is really missing ability to work with multiple strips which you could do with speed factor property. Perhaps for more context you can check #109044 - WIP: VSE: Improve retiming UI - blender - Blender Projects

Retiming tool is still available in alpha builds.

I am not aware of such decision. Currently there were issues with libraries at least on windows, and I didn’t have oportunity to find solution, but I am planning to resolve this in 4.0.

1 Like

Library issues on windows? Tell me more…

I’ve talked about that in PM on blender.chat - I wasn’t able to build the patch due to errors with FFTW library. Not sure if there has been any changes, it’s been a month so I should give it another try. I am quite helpless when it comes to libraries though.

I was going to say that the problem is that FFTW in Blender is only double precision version but, according to #104895 - Library changes for Blender 3.6 - blender - Blender Projects , It looks like the “float version” is included.

I would like to comment several aspects related to sound which appeared lately:

  • GSoC 2023: Improve Waveform drawing speed It’s great to see the advance of this GSoC. It’s never fast enough! And It demonstrates that sound can be improved.
  • Move a sound strip in the VSE, in units smaller that “one frame”. I hadn’t realized that problem until one user mentioned it. One of the only reasons to keep sound in VSE is the “perfect synchronization”. A clap, a smash, a bomb,… really need perfect synchronization. And it looks only an UI problem, because the “Sequence” internal struct uses float to measure the position, so it supports 0.22 frames, for example
  • At my work, all our video files in the archive have 8 audio tracks, and not always the first audio track is the “right one”. I thought that we were the only ones in the world, but recently another Blender user complained that he had files with several audio tracks and he wanted to select one or other track. At the Audaspace source code, it can deal with the different tracks of a file, but everything is hardcoded in Blender and the Audaspace interface in order to simplify and select the first and only one audio track. What a pity!
1 Like

On @pistolario’s #3 bullet point from the VSE chat:

Reading this: How to import a video file with multiple tracks? - Blender Stack Exchange

At some point, it should be considered to add an option(to the File Browser) to, during import, split audio files with multiple audio channels into individual strips with correctly panned positions.

Under Source in the sidebar there could be a channel selector. And Pan function should have a more intuitive interface: ⚙ D12114 VSE Surround: Expose Pan Presets in Sound Strips menu.

Another thing to consider, in relation to waveform drawing (bullet point #1), if the waveform drawing get optimized enough, maybe it could be considered to add waveform thumbnails in the File Browser and Asset Browser?

(Btw. looking at the VFX & Video module meeting resumes, I wonder why the VSE is even in that module, since none of the meeting resumes displays any interest in the state of and development of the VSE. To me, yet again, this simply stresses the fact that the VSE lacks leadership).

1 Like

By surprise, the patch Graphical Sound EQ has been included in the Main Blender. Thanks to the efforts of @iss .
Now, it would be time to agree to the best suitable GUI for the modifier.

I think that modifying the python code, in order to let the user change the ranges of the graph would be enough to start.


yes, and what happens if there is intersections in ranges and how can intersections be beneficial? this must be discussed

@pistolario No need to thank me. I want good code to be merged.

As far as GUI goes, I would say, that first step would be to make scale logarithmic and draw a grid with values, so user can interpret whole curve at once. Then perhaps custom range may make sense, but IMO would be quite unnecessary.

In any case, that is my opinion. I would probably make design first and have it signed off.

Do you want to continue this development?

1 Like

Hello @iss,
I would be glad to continue the “adventure”. I can be a little slow but I can try :wink:

But this time I / we have to learn from my / our mistakes.

The plan could be:

  • list of requirements
  • mockup the GUI
  • approved by devs in BF?
  • patches needed
  • integrate simple

From my selfish point of view, I would need one person in BF devs that could be involved from the beginning, in order to communicate, ask, say “no” when it is needed, suggest changes,… Could be @iss ?

As a first list of requirements:

  • X scale in log
  • grid under the curvegraph
  • change the ranges of the curvegraph?

Are you planning to create or design the GUI yourself? If so you could join the next VFX Meeting to discuss your plan.

1 Like

The VFX & Video module seems to be limited to “Any contributor working on rendering”

That’s probably a copy-paste error

That’s nice to hear, I agree that things would have to change to make this work

I agree with this. IMO you and @fsiddi should have a chat to exchange what both of you envise and what should be worked on. Then formalize this in a design task. That’s pretty much standard process. Francesco is fairly busy, so sometimes he responds quickly sometimes not. I was thinking about regular meetings, even once a month, but not sure if that would be helpful for me. Perhaps for you it would work better.

Will have to leave coordination up to you two. I am happy to participate in conversations and reviews of course.

Hi @pistolario feel free to write up a document describing (as visually as possible) the reasoning behind the feature and the expected UI/UX. Should fit in one A4 page or less!

Sorry for the delay, but I have been thinking about the subject, and getting information.

I’ve started a Proposition for a confortable UI for the User.
There are 2 main changes:

  • 2 curves on the screen
  • X scale logarithmic

I would be very grateful if other people could comment, suggest,… changes, errors, …


Following suggestions by users, I’ve “simplified” the proposal.
In this case, it would be only one graph with horizontal scollbar and X axis logarithmic

Hi @pistolario
I’ve been following this thread for some time because I use EQs with Blender externally– synced with Ardour via Jack. It is so wonderful to see the VSE slowly entering the DAW world :slight_smile:
A few remarks on what’s currently available in Blender 4.1, trying it for the first time and probably biased by my previous experience with digital and analog parametric EQs.
Graphic EQs don’t control bandwidth by definition afaik (only gain and frequency). This module allows bandwidth control, so I would call it Parametric EQ :slight_smile: But the way it does so tends to create asymmetric curves, while the most common use for filtering frequencies is to start with a curve that is centered and this is quite difficult to do with the current design. The center frequency is also very important as a reference point, more than the start/end frequencies. It determines how audio filters are measured in general.
Also, I miss a lot the ability to see and control the bandwidth (or Q) of a curve numerically. Filter shape buttons are rather confusing to me honestly.

A direction I can imagine could be that the module allows only one curve and the panel is limited to three dots (start, center, end). This shouldn’t be a limitation for creating complex filters since modules can be stacked already. The panel would then begin showing three-dots at 0dB, representing a ‘muted’ predefined curve. The middle dot always sticking to the center unless de-centered moving the dots from the sides. In the values section I would add: center frequency, bandwidth and gain.

This is roughly how I can imagine it, but I don’t want step over all the efforts already done. Just hoping that it helps somehow, a bit of a standard view. I’d be happy to collaborate on the usability side if opinions are still needed, I can offer my passion and experience using and designing (analog) PEQs. Also if chat or live sessions are happening.

–Other ideal improvements could be increasing responsiveness, because audio is muted for quite a while on filter changes or enable/disable actions. Or showing the realtime waveform on the panel’s background, but that’s luxury I guess and too visual for an audio filter some could argue.

PD. It’s true that the linear x-axis makes it extremely difficult to use also (see picture) but that has been noted already.

Great work! A light in the dark really.



This is the next version of the proposal, which I’ve just sent to BF developers.
It’s the same, with some tweaks and some implementation details / doubts.