- User Since
- Feb 4 2009, 8:52 AM (650 w, 4 d)
Thu, Jul 22
My response was based on the assumption that ffmpeg is providing us the necessary information, handles this correctly and we have found the necessary information in its datastructures to handle. That's apparently not the case.
Tue, Jul 20
Good point. It's not exactly user friendly to have these two settings that far apart. Question is if the playback and render setting should somehow affect each other? But that's a different story.
Okay, so I looked more detailed into the code and the sample file you sent.
Sat, Jul 17
I've added myself to the reviewers here.
Thu, Jul 15
Can you please send me a sample file where this issue happens, so that I can try/investigate it?
Mon, Jul 12
Sat, Jul 10
I'll close this as a duplicate of T88906.
Mon, Jul 5
Okay, so you have pulseaudio and it works for other applications. Can you start playback in blender and then check if it shows up in pavucontrol (the pulseaudio volume control application, you might have to install it)? For me it looks like this:
Maybe Blender is assigned to the wrong audio device or so?
Sat, Jul 3
Ok, @Richard Antalik (ISS) doesn't have the issue anymore in the current master. @Stick (StickMeister125) can you try if you still have the problem in the 3.0 alpha daily build: https://builder.blender.org/download/daily/ ?
If you do sound animation in blender, you should make use of the Update Animation Cache button that you can find under Scene Properties -> Audio frequently.
Fri, Jul 2
Yes, PulseAudio has been added as a new backend, because OpenAL has other issues (3D sound being different in playback and rendering), so it will sooner or later be dropped completely. But before that we have to iron out the bugs of the new PulseAudio backend. Any help to test it (e.g. with the build above) is appreciated!
Sure! Unfortunately, I didn't get any responses yet from the issue in the pulseaudio tracker. Here's the link: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/issues/1221
I close this as a duplicate of T88089.
I haven't heard of this distribution before. Does it even run PulseAudio? Could you try e.g. pactl list as command in a terminal and see if it outputs something? If that works, you can also try playing something with paplay <filename> where filename should be the path to an audio file (wav or ogg ideally, mp3 doesn't work, at least for me) and check if that works.
Thu, Jul 1
@Stanley (AMixOfGeekStuff) could you try if this build works with pulseaudio (!) on your system: https://builder.blender.org/download/experimental/blender-3.0.0-alpha+master.a112adf16a06-linux.x86_64-release.tar.xz
Not sure, could be. My computer is too weak to even try the file properly, even without scrubbing first, the FPS is way too low on my system... Please make the experimental build @Richard Antalik (ISS)! You may also try this build here which has some recent fixes, but I feat it won't solve the issue: https://builder.blender.org/download/experimental/blender-3.0.0-alpha+master.a112adf16a06-linux.x86_64-release.tar.xz
I just committed bugfixes for pulseaudio and I think that this particular issue could be solved there, could you try this build please (with Audio Device set to "Pulseaudio"): https://builder.blender.org/download/experimental/blender-3.0.0-alpha+master.a112adf16a06-linux.x86_64-release.tar.xz
I'm closing this for now, since there's been no response anymore regarding this and @Falk David (filedescriptor) doesn't have the problem anymore. Feel free to reopen, if the issue appears again!
It's a hack, but this behavior is still weird. Can you try D11778 and check if that solves the issue better? That's a more proper fix.
Great, thanks for the info! Since this is a configuration issue that happens with the default configuration, we'll have to keep this in mind, but as long as there's only a single person having the issue, we don't have to change the default configuration I think. Closing for now!
Jun 14 2021
Ok, I can't really reproduce this on my setup here, but I have some ideas.
Great! Seems like our default is too low for your system. What is the lowest value that works fine?
@Richard Antalik (ISS) can you try if this patch fixes the issue?
What happens, when you try WASAPI with a bigger mixing buffer size? Does it work better then?
Jun 8 2021
It's unclear what exactly the problem is, whether it's in the way Blender uses pulseaudio or if the issue is within pulseaudio. As a workaround for now, you can switch the audio device in the preferences under system to OpenAL or SDL.
Thank you, I created an issue on the pulseuaudio issue tracker with your log.
I'll look into it!
I'll look into it!
Likely a duplicate of T88089. What distribution are you using? Debian by any chance? Also, what is your pulseaudio version?
May 27 2021
Unfortunately, I and a pulseaudio developer I talked to cannot reproduce the issue, but there are some things we can try:
Apr 22 2021
Apr 15 2021
Apr 9 2021
Can't reproduce the issue. Please provide a simple blend file where it happens, as @Ankit Meel (ankitm) wrote.
Apr 8 2021
Mar 29 2021
Mar 28 2021
I can add some documentation that the functions are only to be called when the device is locked.
Mar 27 2021
Mar 19 2021
Mar 17 2021
This was certainly not @Ankit Meel (ankitm)'s fault. I was pushing these changes to get them in within bcon1, together with the other new audio backends, in order to have native support on all platforms.
Mar 16 2021
There was a threading issue, it's fixed now!
Added dynamic loading of PulseAudio.
Mar 15 2021
I added CoreAudioClock functionality for hopefully better A/V synchronization. Please test it! :)
Mar 14 2021
Mar 13 2021
This update fixes a minor issue, where playback wouldn't start properly due to a wrong buffer size.
Mar 12 2021
Mar 11 2021
I've updated the implementation for the case where the API changes the pointer (not following the documentation here: https://docs.microsoft.com/en-us/windows/win32/api/audioclient/nf-audioclient-iaudioclient-isformatsupported).
Mar 5 2021
Mar 2 2021
Feb 5 2021
I guess you mean T55366? And no, none of those have been addressed. They are all limitations that have been there from the beginning and not really bugs, thus more like feature requests.
Please read the other comments - this is not a bug, but a feature request, so there's nothing to be fixed. And if you now wonder if this feature has been added yet: no.
Jan 29 2021
Correct, the NLA strip is only there to signal the start of the playback of the sound. I remember when I implemented this: I wanted a user interface similar to the dope sheet, where you could simply set key frames in which the sound starts to play, but that was not possible. Instead I used the NLA editor, where the strips have lengths... So this was actually a conscious decision to have this behaviour!
Dec 24 2020
That one was a bit harder to find. Running down the wrong rabbit hole, I also found another inaccuracy! :D
Dec 8 2020
Correct, the blender_config.cmake, does not exist in upstream audaspace.
Nov 4 2020
@Clint Hobson (Domarius) that just means that blender is the only application that uses OpenAL on your system. I'm quite confident that it is the ~/.alsoftrc file, you can ensure by removing the file and restart blender and then put it back without ever switching to SDL.
Nov 3 2020
@Philipp Oeser (lichtwerk) I don't have a Mac, so I cannot really be aware of the issue :-/
Sep 7 2020
I made the according changes in rB9cac181fbead. The mixdown functions now expect two more parameters, the first is a function pointer to a function that gets a float and a void*, where the float is the progress (0-1 range) and the second parameters is the second new parameter, i.e. the void* is passed through.
Sep 3 2020
Would it suffice if I add the possibility to add a callback that is periodically called during the mixdown function call? If so, how often should it be called? Without further measures it would be probably be called multiple times per second (i.e. calling it every iteration of the mixdown for loop). Alternatively I could call it every N samples, or check the system time and call it every N milliseconds.