Navigation

• Main page • Troubleshooting • Improving performance • Recording for YouTube • Recording Steam games • Recording game audio • Live streaming • Custom codec options • Translations

• Offline version

Log in

No account yet? Register!

Last modified: Wed, 10 Jun 2020
Refresh

Recording game audio

Using ALSA
Using PulseAudio
- Recording only game audio
- Recording game audio and microphone at the same time
  - The easy way
  - The hard way
Using JACK

Sometimes you don't want to record your microphone, but the audio from the game you're playing. There are two ways to do this: either through PulseAudio or through JACK.

PulseAudio is enabled by default on Ubuntu (except Lubuntu) and many other Linux distributions. PulseAudio works, but is sometimes unreliable and it still has some bugs that you may run into when you try to do more advanced things with it (such as the commands needed to record a game and a microphone at the same time). If you care about reliability and sound quality, I recommend using JACK instead. JACK is harder to set up, but I think it's worth it.

Using ALSA

To my knowledge, there is no reliable way to record the speakers in ALSA except by physically connecting the line out/headphone jack to the line in/microphone jack with a cable. It's primitive but it works very well, you just have to adjust the recording level in 'alsamixer' to get the volume right. You should also disable 'auto mute' if your speakers become muted when you plug in the cable.

Using PulseAudio

Recording only game audio

Assuming you have PulseAudio, you can follow these steps:

Run SimpleScreenRecorder.
Make sure that the audio backend is set to PulseAudio.
Select an audio source that starts with 'Monitor of'.

If you want to switch back, just select an audio source that is not a monitor.

Recording game audio and microphone at the same time

This method is know to be buggy and will likely result in some stuttering. Consider using JACK instead.

Credit goes to Joshua Hertlein for figuring this out. There are two ways to do this: the easy one and the hard one.

Hint

Before you do any of this, I highly recommend that you set the PulseAudio setting 'flat-volumes' to 'no', otherwise you won't be able to control the volume of applications individually (or at least not in a sane way). This seems to be the default setting in Ubuntu, but not in Arch Linux, so you may want to check this. To disable 'flat-volumes', add/modify this line in /etc/pulse/daemon.conf

flat-volumes = no

There is also a bug in PulseAudio that is triggered when PulseAudio switches between 44.1kHz and 48kHz audio. The fix should be released with the next version of PulseAudio (probably version 4.1). Until then, add/modify these lines in /etc/pulse/daemon.conf to work around this bug:

default-sample-rate = 48000
alternate-sample-rate = 48000

This will force PulseAudio to use only 48kHz (everything else is resampled as usual). This may increase CPU usage and reduce the audio quality, but the difference is very small.

After changing these settings, you should reboot so they can take effect (or just restart PulseAudio, but that will probably break your volume control).

The easy way

Tell PulseAudio to send audio coming from the microphone to the speakers, and record the speakers with SimpleScreenRecorder as explained above. The downside is that you will constantly hear your own voice which can be a bit distracting, and you should also set your speaker volume really low (or just use headphones) to avoid echo. But it is simple. Just run this command:

pactl load-module module-loopback

And when you're tired of listening to yourself, you can stop it again with this command:

pactl unload-module module-loopback

Note that the volume of the recording does not depend on your speaker volume. So even when you think the volume of the game sound is really low, it can actually be really loud in the final recording. You should reduce the volume of the game either in the game itself or with PulseAudio Volume Control, so your voice will still be audible. The preview in SimpleScreenRecorder can help you here, it shows the actual volume of the recording.

The hard way

Create null outputs and connect them with loopback streams so the game audio is sent to the speakers, but also combined with the microphone and sent to SimpleScreenRecorder. This is what we want to achieve:

Image: pulseaudio-loopback-diagram.png

To do this, run the following commands:

pactl load-module module-null-sink sink_name=duplex_out sink_properties=device.description=duplex_out
pactl load-module module-null-sink sink_name=game_out sink_properties=device.description=game_out
pactl load-module module-loopback source=game_out.monitor
pactl load-module module-loopback source=game_out.monitor sink=duplex_out
pactl load-module module-loopback sink=duplex_out

Now you should open PulseAudio Volume Control and send the audio from the game to the second null output (game_out), and set the audio source in SimpleScreenRecorder to the first null output (duplex_out). This should be done from the input page - do not use PulseAudio Volume Control to change the audio source for SimpleScreenRecorder because the change will be lost whenever you pause and resume the recording.

As with the previous method, you should probably reduce the volume of the game a bit, because it's usually very loud compared to the microphone volume.

When you're done, remove all the extra outputs and loopback streams with this command:

pactl unload-module module-loopback
pactl unload-module module-null-sink

If you use this method a lot, then instead of running these commands manually, you can also configure PulseAudio to execute these commands automatically on startup. To do this, add the following to /etc/pulse/default.pa:

load-module module-null-sink sink_name=duplex_out sink_properties=device.description=duplex_out
load-module module-null-sink sink_name=game_out sink_properties=device.description=game_out
load-module module-loopback source=game_out.monitor
load-module module-loopback source=game_out.monitor sink=duplex_out
load-module module-loopback sink=duplex_out

Using JACK

Unlike PulseAudio, JACK must be configured manually, and while you're using it, you won't be able to use your normal audio system - everything has to go through JACK. So before you start, you should close all programs that could be using the sound system (this includes web pages with sound or video, often even when they are paused).

I recommend uninstalling PulseAudio if you are going to use JACK, because it will interfere with JACK (even when the PulseAudio daemon is not running).

Hint

JACK can be used with a normal ('generic') Linux kernel, but these kernels are not optimized for low latency. As a result, you will have to use a very large buffer to avoid xruns (stuttering). I recommend using a preemptive kernel or even a realtime kernel instead.

Linux distributions that are meant for audio processing will include a preemptive or realtime kernel by default, so there is nothing you should do. Arch Linux also comes with a preemptive kernel. On Ubuntu (and derivatives like Linux Mint), you can install a preemptive kernel with the following command:

sudo apt-get install linux-image-lowlatency linux-headers-lowlatency

After installing these packages, the preemptive ('lowlatency') kernel should show up in Grub when you restart the computer (it should be the default).

You should be able to get the latency down to 20ms or less with a preemptive kernel. You can go even further with a realtime kernel, but there is usually no need for that and it introduces unnecessary overhead.

Before you start, it's a good idea to configure your system to allow real-time threads. If you are using one of the Linux distributions meant for audio processing, this is probably already the default. If not, you have to edit the file /etc/security/limits.conf and add/edit these lines:

@audio          -       rtprio          80
@audio          -       memlock         unlimited

This tells the kernel that users in the audio group can create real time threads and lock memory to RAM. These settings will make xruns less likely, so you can use lower latencies. It is not required but highly recommended.

Next you have to make sure that you are actually in the audio group. In a terminal, run the groups command to see what groups you are in. If it includes 'audio', you are fine. If not, run this command:

sudo usermod -a -G audio yourusername

This adds you to the audio group. You have to log out and log in again before it will take effect.

Before you can use JACK, you have to start the JACK server. The easiest way to do this is with the QJackCtl GUI, which you can install from the repositories. If you have a choice, I also recommend installing jack2 instead of jack1 - they are compatible, but jack2 performs better on multi-core systems. On Ubuntu, you can install all this with the following command:

sudo apt-get install qjackctl jackd2

Now open QJackCtl, and before you do anything else, click the settings button. Most of the default settings are fine, you just have to make a few changes:

Make sure that the 'realtime' checkbox is checked, and the 'no memory lock' checkbox is unchecked.
Set the same rate to 44100, because this is what SimpleScreenRecorder will use. Other sample rates will still work, but it's not very useful since SimpleScreenRecorder will resample it to 44100 anyway.
Set the period size to a reasonable value, for example 512 or 1024. Lower values will decrease the latency but increase the probability that you will get xruns (clicks or short interruptions in the audio). The best value highly depends on your sound hardware and how well the Linux drivers for that hardware perform. USB sound cards usually perform better here. With a special realtime kernel and the right hardware, you can get the latency down to a millisecond, but I don't recommend actually doing that: it introduces a lot of overhead, could cause xruns, and you probably won't be able to notice the difference anyway once the latency is 10ms or less.
Set the number of periods to 2. If that doesn't work properly, even with very large period sizes (i.e. you still get xruns), try 3 or 4 instead.
Set the number of input and output channels to 2, because SimpleScreenRecorder currently only supports stereo.

Once you've chosen your settings, click the 'start' button and wait a few seconds. If everything goes well, the status display should light up and contain only green and yellow text. If things go wrong (you get xruns), the text will turn red.

You should always make sure that JACK is running before you start any application that tries to use JACK, because they may try to start JACK themselves (in the background) which will confuse QJackCtl (if this happens, you have to find the jackd process and kill it manually).

You can test JACK with any application that has JACK support (for example Audacity). Most normal applications don't have JACK support though, so you will need what's called an ALSA-to-JACK bridge. There are two ways to do this. The classic method uses an ALSA loop device. It always works, but it adds a lot of latency which is annoying. You can read how to do it here, but I won't go into it because there is a much better solution: the JACK plugin for ALSA.

In order to use the JACK plugin for ALSA, you will have to create a file called .asoundrc in your home directory, with the following content:

# override default
pcm.!default {
    type plug
    slave {
        pcm "rjack"
    }
}

# ALSA-to-JACK bridge
pcm.jack {
    type plug
    slave {
        pcm "rjack"
    }
    hint {
        description "JACK bridge"
    }
}
pcm.rjack {
    type jack
    playback_ports {
        0 "system:playback_1"
        1 "system:playback_2"
    }
    capture_ports {
        0 "system:capture_1"
        1 "system:capture_2"
    }
}

This will make the JACK plugin the default ALSA device. Most applications will simply use the default, so this is usually all you have to do. Some applications may have a way to select the sound input/output within the application, in that case you should either use 'default' or manually select 'jack_system'.

You can now use SimpleScreenRecorder with the JACK backend. SimpleScreenRecorder can automatically make connections for common use cases (recording the microphone and/or the speakers), but if you need more fine-tuned control, you can manually manage connections with QJackCtl by clicking the 'connect' button (not to be confused with the 'patchbay' button, which is used to automatically apply predefined connections).

You can also combine SimpleScreenRecorder with various other JACK applications to apply audio effects in real-time.

When you're done with JACK, you should first close all applications that use the sound system, then delete the .asoundrc file in your home folder (or just rename it so you can re-use it later), and then stop JACK through QJackCtl.

Comments

Varaz

Comment #1: Wed, 16 Oct 2013, 13:21 (GMT+1, DST)

Maarten Baert's website

Navigation

Log in

Recording game audio

Using ALSA

Using PulseAudio

Recording only game audio

Recording game audio and microphone at the same time

Hint

The easy way

The hard way

Using JACK

Hint

Comments

Varaz

Maarten Baert

Feles

Maarten Baert

Joozey

Bazukas

Maarten Baert

Ubuntuaddicted

Maarten Baert

Xage

Ubuntuaddicted

Cjcwj12

Maarten Baert

Manuel

Ubuntuaddicted

Manuel

Maarten Baert

Dubslow

Ubuntuaddicted

Andabata

Mrknightofthecheese

Mrknightofthecheese

Maarten Baert

Ubuntuaddicted

Mrknightofthecheese

Mrknightofthecheese

Maarten Baert

Mrknightofthecheese

Maarten Baert

Mrknightofthecheese

Cjcwj12

Maarten Baert

Kovacsa

Kovacsa

Kovacsa

Maarten Baert

Mrknightofthecheese

Maarten Baert

Mrknightofthecheese

Maarten Baert

Mrknightofthecheese

Maarten Baert

Thekiwi5000

Maarten Baert

Kovacsa

Maarten Baert

Kovacsa

Maarten Baert

Ubuntuaddicted

Kovacsa

Kovacsa

Tom_b

Maarten Baert

Rkaid

Maarten Baert

Ucenna

Maarten Baert

Ucenna

Shaneroach

Maarten Baert

Shaneroach

Funkyg4mes

Shaneroach

Maarten Baert

Shaneroach

Maarten Baert