ramagottfried
commented
7 years ago My sense is that this is probably either related to the Dante setup or that your Mac mini is not powerful enough to handle the processing. If you can reproduce the problem without the Dante, definitely let us know and we can take a closer look.
The more channels you have the higher the CPU usage will be, so you'll always need to benchmark your patches/live sessions to make sure you're computer can handle it. Remember that you should always test your system well in advance of a show with the actual hardware you are going to use!
This system should be quite capable of being expanded to much higher channel counts. I used something similar with a 75+ speaker MADI/Ethersound based HOA system at Ircam with no problems. However, I was using a quad-core MacPro tower (circa 2012).
The new version with the fx rack will probably end up being more expensive CPU-wise as people start adding more spatial effects, so this issue of managing computer power will be one to keep an eye on.
cameronangeli
From @cameronangeli
No specific bugs, but we're having a degradation of signal. It's either a clocking issue or the encoder can't handle the amount of info we're sending and translating to our speaker arrangements. I'm leaning towards an encoder issue, because we're seeing it regardless of whether we are on one or two machines.
Here's the arrangement we used:
1) Performance Machine (Macbook Pro) running Ableton Live 9 64 bit: We are using all 16 envelop audio sources: 1-4 = 2 stereo pairs that are used as the core, 4 corner virtual anchors placed at 0 elevation in the corners. Drums and bass piped to these anchors so you hear them from all ground level speakers; 5-8 = 2 stereo pairs that are used for send effects piped into static anchors placed in the roof in a square - Reaktor effects like a 4 ch rotate delay, or 8 ch random delay tap, or front to back reverb; 9-14 = 3 stereo pairs for individual stereo sounds that can be panned; 15 & 16 = mono sounds that can be panned.
This machine is networked using ethernet on a static IP address (192.168.50.12 - or .13 or .14 depending on who is performing) to the encoder machine, which also has a static IP address (192.168.50.20).
All envelop M4L plugins are set to connect to the IP address: 192.168.50.20.
We are using Dante Virtual Sound Card 32x32 48K to send audio from ableton digitally over ethernet. So the UDP messages, and 16 channels of audio are sent from the performance machine, to the:
2) Encoding Machine (mac mini 2016 8GB Ram). We are also running Dante Virtual Sound Card, and Dante Controller on this machine. The channels from the performance machine are mapped to the input channels of the encoding machine. In order for the signal to work properly, we also had to initially use Dante Via to identify audio sources at Dante capable.
Max 7 is running on this machine, which is running our adapted version of Envelop. We have changed the IP address for the client and server to be 192.168.50.20. We have created 11 speaker presets - which includes the Savage arena arrangement.
That arrangement is 16 speakers = 4 JBL powered subs at 0 degrees in the corners, 4 3-way Mackie stacks at 2 degrees in the corners, 4 JBL eon 515XTs at about 10 degrees in the corners, and 4 JBL eon 515xts that are mounted at about 50 degrees, rotated 45 degrees from the corners to fill the gaps and add height.
The Mac Mini is connected to a PreSonus 16ch audio interface. This only has 8 analog outs, so we have expanded using ADAT to a behringer 8+8 unit, giving us the other 8 outs.
When it's working, the panning sounds amazing. however, at mostly random times, the sounds will chunk up. It seems like there is channel bleed, and effects propagating through the glitch. On my laptop, the degradation is not too terrible, and lasts for a few seconds when it happens, then fixes itself. But on my buddy's machine, it will progressively get worse to the point where audio stops. We will restart Max on the encoding machine, and all is solved. It also appears that the more we throw at the encoder, the more we notice this random degradation. for example, using the send effects, which will add audio to 4 channels all at once will instigate the issue.
the encoder machine is running at about 15-20% DSP. and the laptops ableton CPU usage never goes above 15%.
We have added a few other things to the Server to help us manage the setup: 1) we added a master volume and meter that allows us to control the volume of all 16 tracks before they enter the encoder. 2) We added sub groups to control the volume of each of our speaker types (like a fader for the subs, fader for the 3-ways, fader for the low eons, and fader for the roof eons) 3) in order to deal with the speaker placement distance discrepancy, we added individual delays to each speaker. We never used this because it sounded fine, but technically, we should have delayed the eons on the room, since they were closer to the DJ than the corner speakers. 4) we disabled the built in reverb because it uses up a bunch of the DSP, and really doesn't add too much to the experience. We prefer to have the effects routed as we did with Reaktor.
We also have two other small scale setups - one in my studio with 8 speakers in a cub arrangement, and one in my buddy's studio with 8 speakers, 4 in the corners and 4 up high rotated 45 degrees. The planning seems to translate very well between all three setups and the binaural option.
There are a few tests that we need to run to further identify why we might be having this issue: 1) we need to do a full analog routing (no dante). based on the gear I have, I could test up to 16 channels out of ableton, to an 8 channel speaker arrangement in my studio. to try the 16 channel setup would require me to find a big enough space. 2) i'd like to get an external clock to sync everything to.
I know this is a lot, and i assume we have taken the envelop software to places that you did not intend, but i thought you could use the feedback to help make this next iteration as good as can be.