| Git Clone URL: | https://aur.archlinux.org/gpu-screen-recorder.git (read-only, click to copy) |
|---|---|
| Package Base: | gpu-screen-recorder |
| Description: | A shadowplay-like screen recorder for Linux. The fastest screen recorder for Linux |
| Upstream URL: | https://git.dec05eba.com/gpu-screen-recorder |
| Licenses: | GPL-3.0-only |
| Submitter: | DEC05EBA |
| Maintainer: | DEC05EBA |
| Last Packager: | DEC05EBA |
| Votes: | 52 |
| Popularity: | 2.99 |
| First Submitted: | 2024-11-15 18:38 (UTC) |
| Last Updated: | 2025-04-15 01:35 (UTC) |
hey, I'm facing an issue, I have no audio in the recordings, using pulseaudio. I'm passing this option -a "$(pactl get-default-sink).monitor"
and this is its output: alsa_output.pci-0000_00_1f.3.analog-stereo which seems right, also I can see from pavucontrol that gpu-screen-recorder is capturing audio properly, as you can see here: https://i.imgur.com/9GgyM0D.mp4
so I don't think it's an issue in my setup.
@Typology thanks for reporting. I think it should be fixed now. channel_layout is deprecated in new ffmpeg so it will be removed by ffmpeg devs in the future. I added a version check to use the deprecated code. I dont have ubuntu lts so I didn't test it properly, but I believe it should work.
hey, current latest branch cannot be compiled with ubuntu LTS 22.04 (that includes popos and many other ubuntu-based distros)
that's because ffmpeg (and therefore libavcodec) are on 4.4 still, which uses channel_layout for audio channels, instead of ch_layout used in 5.0+, so that will produce an error like this when compiling:
src/main.cpp: In function ‘AVCodecContext* create_audio_codec_context(AVFormatContext*, int)’:
src/main.cpp:390:47: error: ‘AVCodecContext’ {aka ‘struct AVCodecContext’} has no member named ‘ch_layout’; did you mean ‘channel_layout’?
390 | av_channel_layout_default(&codec_context->ch_layout, 2);
| ^~~~~~~~~
| channel_layout
src/main.cpp:390:5: error: ‘av_channel_layout_default’ was not declared in this scope
390 | av_channel_layout_default(&codec_context->ch_layout, 2);
| ^~~~~~~~~~~~~~~~~~~~~~~~~
src/main.cpp: In function ‘AVFrame* open_audio(AVCodecContext*)’:
src/main.cpp:516:36: error: ‘AVFrame’ {aka ‘struct AVFrame’} has no member named ‘ch_layout’; did you mean ‘channel_layout’?
516 | av_channel_layout_copy(&frame->ch_layout, &audio_codec_context->ch_layout);
| ^~~~~~~~~
| channel_layout
src/main.cpp:516:69: error: ‘AVCodecContext’ {aka ‘struct AVCodecContext’} has no member named ‘ch_layout’; did you mean ‘channel_layout’?
516 | channel_layout_copy(&frame->ch_layout, &audio_codec_context->ch_layout);
| ^~~~~~~~~
| channel_layout
src/main.cpp:516:5: error: ‘av_channel_layout_copy’ was not declared in this scope
516 | av_channel_layout_copy(&frame->ch_layout, &audio_codec_context->ch_layout);
| ^~~~~~~~~~~~~~~~~~~~~~
src/main.cpp: In function ‘int main(int, char**)’:
src/main.cpp:1192:125: error: ‘AVCodecContext’ {aka ‘struct AVCodecContext’} has no member named ‘ch_layout’; did you mean ‘channel_layout’?
1192 | device, audio_tracks.back().input_name, audio_codec_context->ch_layout.nb_channels, audio_codec_context->frame_size) != 0) {
| ^~~~~~~~~
| channel_layout
any ideas how can this be resolved?
@ThatOneSeong thanks. Also thanks for testing on pipewire. It was a temporary fix for another problem but I think it should be properly fixed now. It works on both pulseaudio and pipewire now.
The CFR change is a pleasant surprise, and I can see getting a lot of use from multiple audio tracks as well! Thank you for your continued work!
FWIW, it seems that PipeWire 0.3.57 has caused specifically the audio track to crackle and desync every couple tens of seconds in a recording. Reverting to PW 0.3.56 resolves it, but not sure whether it's this or that which causes these symptoms
The recorded video now uses constant frame rate and should work fine with kdenlive, vlc and other software.
@Typology thanks, but the videos are still in variable frame rate format which means some video editors (kdenlive) and some video players (vlc) cant handle them correctly (works with blender, mpv and youtube though). I need to make them constant frame rate videos first before the videos work well everywhere :<
Hi @DEC05EBA , I saw the latest commit and saw that you added support for multiple audio inputs with separate audio tracks! amazing work, it's now officially the best screen capturing software for linux, it's even better than the original shadowplay on windows, I didn't know where to write this, but I wanted to thank you for all your work.
@DEC05EBA Can confirm that it's working great now with today's update. I've closed the issue on pipewire to reflect this. Seems odd that a change to better reflect Pulse behavior actually caused more quirks, but am glad to have the solution regardless. Thank you for the insight.
@ThatOneSeong thanks, it should work now (im now setting the fragmentation value manually in the program). Pulseaudio seems to behave a bit different than pipewire with fragmentation size. Even though pulseaudio says it sends a batch every 1.5 second by default, that doesn't actually seem to be the case. It sends data at the requested sample rate without having to set fragmentation value (this is when using pa_simple_new interface).
Pinned Comments
DEC05EBA commented on 2024-11-26 18:52 (UTC)
The aur package gpu-screen-recorder-git was renamed to gpu-screen-recorder. Install gpu-screen-recorder instead or you may experience issues.