High-Frequency buildup at 22kHz with muted AudioTracks

Something weird going on. It is probably a high-frequency buildup at 22kHz, but it can be tricky to see the frequency, because it depends on how you set up the Voxengo Span (Free) in the Edit section, but you can definitely see the peak and rms value going up easily.

1. This is how to see the peak/rms values to change:

-Starting an Empty Project.

-Using the Voxengo Span (Free) on the Stereo Out, hitting START/PLAY ->
Span shows the basic -200dB peak/rms readout.

-Adding Empty Audio Tracks, hitting START/PLAY ->
Span shows the basic -200dB peak/rms readout.

-Filling the 1st Empty Audio Track with a Wav audio file, then MUTING IT and hitting START/PLAY ->
Span still shows the -200dB readout.

-Filling the 2nd Empty Audio Track with a Wav audio file, then MUTING IT and hitting START/PLAY ->
Span now shows a -194dB readout.

-Filling the 3rd Empty Audio Track with a Wav audio file, then MUTING IT and hitting START/PLAY ->
Span now shows a -190.5dB readout.

-Filling 10 Empty Audio Tracks with a Wav audio file, then MUTING IT and hitting START/PLAY ->
Span now shows a -180.0dB readout.

-The readout is getting bigger with every new Audio Track added.

-You have to hit start/play where the audio track events are. It does not matter if it contains an audio signal or it is just plain silence there in the audio event.

-If you hit start/play where no audio events exist Span goes back to -200dB.

-Note that Span needs 12 seconds to show the signal change when hitting STOP in Cubase. (The time it takes in Cubase to fade out to silence after hitting STOP…)

2. This is how to see the frequency that builds up:

  • As I said, it is tricky to catch that little green line visually. I’m not sure why is that, but you can try if you want.

  • Open Voxengo Span

  • At the top right corner there is Edit. Click on it. It opens the Spectrum Mode Editor.

  • Twist the Range Lo knob knob to -180.

  • Now you can hit Start/Play if you filled the project with a minimum of 20 muted audio tracks, but still keep the Spectrum Mode Editor open.

  • Now when the project is playing twist the knob “Freq Hi” upwards very slowly with holding down shift key on the keyboard until you see the little, thin green vertical line right around 22kHZ.


    -DN-

CPU: Intel Core i7-4771 3.5GHz
Motherboard: Asus Z87 Pro
Memory: Corsair Vengeance 8GB DDR3 1600MHz CL9 Memory
GPU: Using the iGPU of the CPU (HD Graphics)
Audio Interface: Focusrite Saffire Pro 14
Windows 8.1 x64
Cubase 8.5.15 x64
Sample Rate @ 44.1kHz

Here you can see the phenomenon with picture attached of it.

Project is started, it is in PLAY.

20 muted audio tracks, no audio signal going through.

Stereo IN also muted.

Voxengo Span (free) placed on the Stereo Out’s insert slot.
It shows a -174 RMS and -174 Peak information.

Also you can see that high-frequency build-up in Span right next to the mouse pointer where it makes a cross.
The build-up is around 22kHz. (The long red vertical line shows the bandwidth limitation of a 44.1kHz session)

Cubase’s Master meter shows the same -174dB Peak Max information.

When project is stopped, then Span shows -200 RMS/Peak. Cubase’s Master meter shows -infinity.

Can anyone confirm?

-DN-

peak_rms_buildup_no_audio_signal.jpg

22050 is the band limit for a 44.1kHz sample rate.
Does the frequency you see change to 24000 if you change the project rate to 48kHz and perform the same test?

I set up a project with 32 audio tracks. 44.1kHz, 24-bit. I put Span on the stereo master channel as an insert.

Playing back silence (no audio at the cursor position, the Master Meter’s Peak Max value remained at -infinity.
No indication on Span.

Muted all the tracks whilst playing back silence (no audio). The Master Meter’s Peak Max value remained at -infinity.
No indication on Span.

Duplicated an audio clip 32 times. Played back in that region with all channels muted. The Master Meter’s Peak Max value read -172.9dBFS.
Span showed an initial peak at 22kHz up to -156dBFS which then subsided to zero. The Span Peak reading read -172.9dBFS

I saw small peaks at 22kHz on Span every time I started playback, which then reduced to nothing after a couple of seconds.

I then started playback (whilst muted) in the audio region (Peak at -172.9dBFS), let it play on into the blank region for a few seconds, then reset the Master Meter and the Peak Max read -infinity.

I then raised the audio level in the clips by 12dB, but this had no effect on the Peak Max value.

I then converted all the audio clips to silence and played back in the audio region. The Master Meter Peak Max value read -infinity.

I restored the audio and then set all the channel faders to -infinity with Mutes off. The Master Meter read -infinity. but putting the Mutes on again gave me -172.9dBFS.

I then tried to see how many muted audio tracks were required to give me anything but a -infinity reading, and it took 15 muted tracks to get above -infinity at -179.5dBFS Peak Max.

It took 1040 muted audio tracks for the Peak Max to reach -142.7dBFS (The theoretical maximum noise floor of a 24 bit system). Practical noise floors for even the very best audio interfaces are in the -115 to -120dBFS region, and my PC would run out of computing power long before I could play back enough tracks (I estimated at around 10,000) before the noise hit the real world noise floor of the very best audio interfaces.

I exported a mix with 1040 tracks muted in 32-bit FP format, and Wavelab just reported an analysis of the file as all being -infinity, so no measurable noise.

So, from what I’ve seen, muted tracks with audio produce a very, very low level of background noise (so far below the noise floor of any audio playback system that it makes it purely academic), which drops to nothing if there is no audio or sections of silence. I have no idea how the mute function works on a channel, but it doesn’t look like the output is being set to zero (which would seem to be the obvious thing to do), but maybe multiplied by a very very small floating point value (possibly to avoid a divide by zero error). These small numbers do add up, but you’d need a ridiculous number of muted tracks to make any impact on your mix at all.

And whilst I was seeing small peaks at 22kHz on Span, they did vanish and I can only conclude they are either an artefact of the integration period of the display (e.g. it needs a bit of time to gather all the samples over its integration period) or something to do with its integration into Cubase (as it also showed a brief 22kHz peak when not playing back but I simply reset the Cubase meter!) or maybe something to do with its Fast Fourier Analysis on a bandwidth limited signal, or possibly ringing in its low pass filter.

So whilst I’d agree the issue does exist, and possibly could be fixed by Steinberg, it really seems to have no real-world effect. It probably seems worse because of the Span display, which I don’t think shows what is really happening.

This explains a lot… I think this might be connected to the reason Cubase 8.5 sounds so “different”… and not necessarily in a good way. I’d really like to have them take a long second look at the audio engine itself for these kinds of flaws. Some day I’d like to use Cubase 8.5 or 9 but as long as the entire thing is so LOOSE and bugged; and cpu inefficient… it can never replace what a solid workhorse 6.5 has always been.