need speed

Audacity 2.1.2 and Xubuntu 16.04 64 bit. I need to record 18 channels of music via USB cable from a mixer. I have a 1TB spinning drive running the program and a 120G SSD to record it onto. I went into Preferences and chose the SSD drive for the temp file. Is this correct? Is this the fastest means of recording to make sure that I don’t have to drop the quality? My issue has always been trying to get enough throughput to the HD.

I had actually gone backwards before to Xubuntu 14.04 and an older version of Audacity because it worked where the new version did not. But lately, I can no longer edit the recordings made on the older Audacity version on the new version: when I try to play something, it skips ahead about 20 seconds before actually playing.

Thanks in advance.

Set the CPU frequency (clock speed) governor to “performance”.
Give Audacity higher CPU priority (less nice).

Actually, CPU time is not the issue. When I am recording, the CPU is running at about 10 %, memory usage is around 880 meg, and 0 swap. I am running a 3.3 quad core.

What I noticed: with the older Audacity 2.0.?, when I finally stop the recording after about 15-20 minutes of recording, the HD runs crazily for about a minute before I am able to save the file. With the newer version of Audacity, I can save the file instantly. What I am supposing here is that the old version was “stringing” everything together AFTER the recording was stopped whereas the newer version does the stringing while recording. Therefore, I am assuming that this is where the problem lies: the HD is not fast enough to lay down all that info. So, that is why I have put an SSD drive into the machine. I do not yet know whether that is going to do the trick or not. I was just trying to understand exactly how and where Audacity buffers to and records to.

Right now, I have the spinning HD running the program and the SSD recording the soundtrack. I could have the program run and record from the SSD if someone tells me that is a better way. But again, the writing of the soundtrack is where I believe the bottleneck occurs.

This is an older post about this problem: https://forum.audacityteam.org/t/allen-heath-qu32-board-issues/42701/7 If the sound clips are still there, you can hear what I was getting while recording: and that was with the recording quality settings brought down quite a bit with really no difference from what Audacity used as default when installed.

One last thing: a 15 minute recording (on the older version) @ 18 channels, comes out to about a 6 gig file.

I’m awestruck that you can get Audacity to record 18 channels.

Within an Audacity-project the audio is stored at 32-bit-depth, (no matter the bit-depth of the capture).
18 mono channels at 44100Hz 32bit is ~80Mb per minute, for 15 minutes is 2700Mb = 2.7Gb

Maybe you’re recording at higher-rate than the standard 44.1kHz,
96kHz sample rate would give 5.88Gb for 15 minutes, that’s close to your “6 gig”.

18 stereo channels, hence the 6G size. Or there abouts.

Anyway, I am still trying to find out what the best set-up is for attempting it, if someone has any ideas.

I did it before with the older version. Now, I have better PC, better HD (SSD), and I am hoping to accomplish this. The SSD is hooked to a 6G SATA port. I just have no other ideas other than trying to find out exactly where Audacity buffers/records the info.

I have the program on the spinner HD, the temp directory for Audacity on the SSD. I am hoping that this is the correct/best option.

Recording was at 44.1 with 32 bit float. What else can I tweak if I need to? I am hoping not to drop below 44.1 one for several reasons: I want the quality and even when I tried that before it really didn’t help.

No joy today. Even with the SSD, I still could not record 18 channels even after dropping the to frequency to 22.1 and using the lowest sampling rates.

Have you tried using Jack Audio System?
http://jackaudio.org/

I have not. Anytime I have ever tried using JACK, I have never had good luck. However, I have not tried it using Audacity.

I still believe that what I am trying to do is just too much info to be processed and written to disc that quickly. The older version of Audacity wrote to the disc differently than the newer version. What I probably need is a Dante card, but that is not in my budget.

The amount of data being written has not changed much. The vast majority of the data is the 32-bit PCM “AU” data files, then there’s a tiny bit of data in the AUP file. Audacity has been using the same format for AU files for many years.

I’ve just tested recording 20 channels simultaneously using Jack, Audacity 2.1.3, a spinning hdd, an i7 processor laptop with 8 GB RAM.
The waveform updates a bit sluggishly with waveform data appearing about every 3 seconds, but the recording is perfect. If a spinning hdd can handle 20 tracks at 44100 Hz sample rate, 32-bit float format, then it should be no problem at all for you SSD. (I’ve also tested recording onto a SSD and that also works flawlessly, though recording length for 20 tracks is limited by my SSD being quite small capacity).

You do need to avoid on-the-fly resampling while recording a lot of tracks, otherwise you either compromise quality due to low quality conversion, or sacrifice performance due to high quality conversion. Ensure that you use the same sample rate throughout.

You have my interest now. Are you using JACK? What are you changing in the “out-of-the-box” Audacity setup? Realize (and you probably already do) that the stream coming in is purely digital: straight out of a Allen & Heath QU32 mixer. It comes in via a USB cable directly to a USB port on the PC. It does not go to the sound card or any other card.

I do not know how Audacity records the data. What I do know is that when I used the old Audacity (2.0.?), when I hit the STOP button to stop the recording, I would have to wait for 1 to 1.5 minutes before I was able to click on FILE and name and save the file. While waiting this time, the HD light was on solidly. When the light went off, I could save the file. But, with the new Audacity 2.1.2, as soon as I stop the recording, I can instantly hit the FILE icon and save the file. So, there is something different that Audacity is doing in writing / saving the file. It worked before: all 18 channels with no problem, but it doesn’t now. What happens now is that about every 2 minutes, I get this buzzing noise across all channels. It lasts about 20 seconds and then clears itself only to repeat again about 2 minutes later.

I just saw where you said that you did use JACK. Why is that making a difference?

In effect, the QU32 is a (very large) external sound card. Your computer has no idea that it is a mixing desk. All that your computer sees is a multi-channel USB audio device. Your computer knows that it is an audio device rather than some other type of USB device, because ALSA (the audio device driver) tells the Linux kernel that it is an audio device and handles data transfer between the device and the sound system’s application interface.

By default, most modern Linux desktop distributions use PulseAudio as the default sound system. PulseAudio offers many benefits, such as on-the-fly resampling, device sharing, audio stream mixing, and more. However, this does come with a price of processing power.

To use Jack audio system it is necessary to bypass PulseAudio. Jack requires exclusive access to the audio device and everything has to run at the same sample rate (no resampling). Whereas PulseAudio will start and stop audio streams on demand, when you start Jack, it starts its audio streams and runs them continuously. Applications are then ‘patched in’ to those streams. Once the streams are running, there is virtually zero overhead for applications connecting to the audio streams. The efficiency of Jack is comparable to that of ASIO on Windows, with the addition of powerful audio routing capability similar to, but more flexible than Rewire.


When recording into a new project, Audacity writes the audio data into its temp folder. On saving the project, data that is in use by the current state of the project is copied from the temp folder into the project “_data” folder. On closing the project, unused data (the “Undo” history) is deleted.

When recording into a project that has been saved, Audacity writes the audio data directly into the project’s “_data” folder. Saving a project that has previously been saved is very much faster because no data needs to be copied - the data is already in the project “_data” folder. On closing the project, unused data is deleted from the “_data” folder.

Thank you so much for explaining that. Before I (we, if you are willing) go further, can we continue this here or do I need to contact you some other way?

This system is used at my church, so even though I have access to it anytime that I want, I do not have access to guitars, drums, keyboards, and singers except on Sundays and a few other times. So, anything to be tried will usually only happen one day a week. I run the sound room for 2 of 3 services. First service is for making sure that the sermon is recorded and uploaded to the website. With the new version of Audacity, I am doing that via analogue hook-up, which is fine. However, once I have that done, I have the second service to experiment with recording the 18 channels of music.

So, I am assuming that the next thing that I need to do is install JACK. What do I need to specifically do? I kinda remember an awful lot of questions during the install of JACK. I also remember pretty much screwing my system up to where I just reinstalled it.

Again, thank you for your help so far: you have no idea how much it is appreciated.

Here is best for me.

Jack should already be installed unless you built Audacity from the source code without Jack support. Jack is normally a dependency. By default it does not start on log-in, so you need to start it manually (can be scripted later when everything is running properly.

Assuming that you have a fairly standard Xubuntu installation, the default sound system will be PulseAudio, and PulseAudio Volume Control (pavucontrol) should already be installed.
If you run:

sudo apt-get install pavucontrol

it will either tell you that it is already installed, or it will install it. Either way it is helpful to have it installed.

For configuring and starting Jack, install QjackCtl
So that Jack will play nicely with Pulse, install pulseaudio-module-jack

apt-get install qjackctl pulseaudio-module-jack

Then configure qjackctl to run the following command after startup. Copy it into “Setup…” > “Options” > “Execute script after Startup”:

pactl set-default-sink jack_out

IMPORTANT - Ensure that no audio applications are running while you set this up or the set-up will fail.
It’s a good idea to reboot or at least log out and back in again to ensure that all sound services are ‘fresh’. Jack MUST have exclusive access to the audio device.

Now you need to set up the options in QjackCtl so that Jack uses the correct audio device (your USB mixer).
After selecting the input and output devices, try starting the Jack server by clicking the “Start” button in the main interface of QjackCtl. If Jack fails to start (quite likely) you will need to fiddle with the Sample Rate, Frames and Buffers settings to get it to run. Don’t aim for minimum latency - aim for maximum stability.

Once you manage to get Jack to run (without “xruns”) then that’s the hard part done.
Let me know how you get on.

I will keep you updated because I am sure that I will have questions along the way. I will try and get out to the church in the next day or 2 and bring the machine home to play with.

Alright: finally got back to this thing after a very busy week.

I started Jack and put that line in to run after start-up. I opened Audacity and under Preferences set it to use Jack. I opened Jack and have a little window sitting on the desktop. Now, I am lost. Do I have to hit the Play button in Jack or will Audacity handle that for me? How can I test this at home: I want to at least get a feel for what I am doing before hauling it out to church and swapping out the other machine.