If you’d be so kind, tell me more about how the automixer application works.
I am used to using the Shure SCM810’s. I like that fact that only one mic will come on per sound source, even if you put two mics at equal distance to a talker, only one will activate.
Additionally the NOM feature of the Shure is pretty essential for live PA type event where feed back could be a real issue as more mics open. As you probably know, the 810 reduces the overall output by 3db every time the number of active sources is doubled.
Finally I always use the last mic on feature so that if there is silence, you still hear the room ambiance. This is especially important on a lot of the video shoots I do.
Have you been able to address any of these issues with your program?
Thanks for your interest in this program. I wrote it simply as an experiment to “see if it could be done” and it’s based on the Dugan algorithm. That is that the total level stays constant and the level for each channel is set based on the signal level of that channel. It differs from the 810 (from what I’ve heard) in that the 810 is more of a gate than actual automixer.
I’m not suggesting that you necessarily use it for a live event, but that you try evaluating it’s operation to see how it works. It does not have features like last mic on, but I don’t know that the Dugan does either. Supposedly the Dugan is the benchmark for automixers so I based it on it’s operation.
To use it, you simply use the parameter change message function of the LS9 (RX and TX) and connect your computer to the LS9 via MIDI and run the program. You pick your channels (up to 8, they must be together) and then activate it. You can see the operation of the software on the screen.
I tried your automixer software this afternoon. Nicely done and I like that you kept it simple.
I am not sure if I was doing everything correctly. For “testing” purposes I took a tone generator and “Y” cored it into the first four inputs. I set all channel gains the same and brought up channel 1 so my output meters on my LS9 read -10. I then raised the level of channels 2-4 to the same level as channel 1 and observed that the output level on the LS9 was now -4. Your software DID indicate that channels 1-4 gains were being redueced about 6dB each, but the fact that that did not seem to be what was actually happening on the output nof the LS9 concerned me.
Am I misunderstanding your design, or maybe doing something else wrong? Any ideas? I thought all automixers reduced the overall gain as you added “active” channels. I’m worried this potentally could run a system into feedback if a panel discussion got out of hand and everyone started talking at once.
Thanks for trying out the program. I really should put some documentation together for this program!
It operates by adjusting the attenuators on the channel, so it’s pre-fader, not post. Even though many people like to use automixers post-fader, I didn’t like the way post-fade works. I like using the fader to balance the levels AFTER the automixer. But, if you pull a channel fader down, it does take that channel out of the algorithm. I think muting the channel also takes it out of the algorithm.
So, to test it, you’ll need to send signal to one channel, leaving the inputs to the other channels disconnected. Set the gain on each channel the same. Set the faders to a nominal level, and then plug in another channel. The output level should stay the same, I think. It may increase slightly due to the fact that those 2 inputs are the same signal. If signals are different, the level should stay the same from 1-8 inputs.
Thanks for your quick responce and for all your help. Also thanks for the explanation of how it works by adjusting the attenuator. I almost forgot about the attenuator and was wondering how the automix program was altering the levels.
I tried what you suggested with a 700k tone only plugging in channel 1, setting the gains and faders equal on channels 1-4, then one at a time plugging in channels 2-4. I had the same results with the overall output of the LS9 incressing by 6dB.
I then tried doing a similar experiment with a 700k tone into channel 1 and a 1k tone into channel 2. Unfortuately they also summed and gave me an overall increase of 3dB. By the way, I DID take a look at the attenuator screen on the LS9 and it DID show the attenuator kicking in as I added more channels. In fact, when I FIRST began to bring up the second channel, the overall output of the LS9 DID drop, but by the time I raised the fader to the same position as channel 1, there was an increase in the overall output, even though the attenuator seemed to be adjusting accordingly.
This doesn’t make sense to me. Mathmatically and electronically this shouldn’t be happening. Any idea what’s causing this?
It sounds like the calculation for the sum of all inputs is using the ‘10LOG rule’ for power calcs instead of the ‘20LOG rule’ for amplitude/voltage calcs ie: -18dB + -18dB = -15dB instead of -12dB. So for 4 inputs each at -18db the summed level would be -6dB, and according to Dugan’s algorithm each input should be reduced by 12dB, but it sounds like for every doubling of amplitude (6dB) there’s only a 3dB reduction in input gains.
I expect there’s something obvious I’ve missed…
Here’s a few update ideas I had:
More channels!
The ability to assign which pot(s) the automixer acts on. With the channel attenuator if your PC crashes there’s no easy way to get a manual mix back, if you had the option of selecting mix pots instead you could use a mix buss as your automix return, with a non auto mixed buss as an emergency backup.
Pre/Post fade options for channel att automixing.
Even more channels!
Thanks for writing this app and releasing into the wild for free.
We tested it with 20log and 10log and the way it works at the moment seemed to be best when you’re using non-coherent sources, which is usually the case when you have a mic on each person in the discussion.
The forumula is easy enough to change, but you need to test it with live sources, not test tones.
That being said, I have never used the program myself, only Jens Brewer (who made the first post in the AutoMixer forum) has actually tested it. John Roberts, who used to manufacture automixers verified the forumla. From his tests, it seemed to work properly. JR suggested another way of calculating the overall level which I have not tried yet, as I have no beta testers and I pretty much abandoned this project a while ago.
If you have a way of testing it properly and you would like changes to the formula, I’d be happy to take a look at modding the program.
The automix channels MUST be in sequence (the start and number of channels is selectable in the program) and the max. number of channels is 8. No way around those 2 limitations due to the way the Yamaha sends level information.
Using mix levels instead of attenuators is an interesting idea, I can’t remember if there was a reason I did it with attenuators, it’s been a while since I looked at this program. If I could get it to use mix levels instead of attenuators,then also the selection of pre/post would be possible. Interesting.
If you guys think the program has potential and you’re willing to put in the work to scientifically test it, I might be able to find some time to do some modifications. I don’t have an LS9 or M7CL at my disposal, so it could be a time-consuming process as I have no way to test anything that I do.
Thanks for bearing with me on this. I think your program has REAL potential. Inspite of my critisisms the basic principles work and I would be thrilled to test and use this program.
During my testing I observed a number of times where the program locked up. I’m not sure yet what the “sequence” was that seemed to cause it. It may have been as I switched it from active to inactive and back again. Obviously having it “stable” is the most important thing for me. This really is a terrific idea, and I believe it will do what I need. I am going to try to do an event or two that is “unimportant” (maybe a freebie) and see how it goes. Any idea what needs to be done to keep it from locking up?
I think your approach with using the attenuators is just fine, and although I don’t quite understand why the overall output level is inconsistant, as I consider it further, I really don’t think that will be a problem. I sort of wish however when exiting the program that it would return the attenuators to zero. I’d also like to see the attenuation levels as displayed in the program match the “maximum” level set within the program. I noticed that if I set the maximum attentuation to -20dB, that the LS9 will never do more than the 20dB (as can seen within the attenuator screen on the LS9) however the level of reduction shown below each channel of the program may falsely indicate substancially more.
Also, why exactly is it limited to 8 channels?
Thanks again. I really appreciate all your hard work.
Thanks, Tom. It’s nice to do something and have it be appreciated.
During my testing I observed a number of times where the program locked up. I’m not sure yet what the “sequence” was that seemed to cause it. It may have been as I switched it from active to inactive and back again. Obviously having it “stable” is the most important thing for me. This really is a terrific idea, and I believe it will do what I need. I am going to try to do an event or two that is “unimportant” (maybe a freebie) and see how it goes. Any idea what needs to be done to keep it from locking up?
That has not happened in the testing I did. I would assume at this point that we’re dealing with something particular to your PC - the version of windows, the drivers for the MIDI interface, something like that. Off the top of my head, if your computer is typically very stable, then I would suspect the MIDI drivers for your interface. What brand and model of interface are you using? What version of Windows are you running? Is your computer normally very stable?
I think your approach with using the attenuators is just fine, and although I don’t quite understand why the overall output level is inconsistant, as I consider it further, I really don’t think that will be a problem. I sort of wish however when exiting the program that it would return the attenuators to zero.
I believe that it does that when you click the “stop” button, does it not?
I’d also like to see the attenuation levels as displayed in the program match the “maximum” level set within the program. I noticed that if I set the maximum attentuation to -20dB, that the LS9 will never do more than the 20dB (as can seen within the attenuator screen on the LS9) however the level of reduction shown below each channel of the program may falsely indicate substancially more.
I think that’s a good idea, never really got around to doing that.
Also, why exactly is it limited to 8 channels?
It’s because of the speed of the MIDI information being supplied by the console. Above 8 channels, the information is coming too slowly for it to work properly. Limitation of the MIDI spec, I’m afraid!
Thanks again. I really appreciate all your hard work.
NP. Like I say, if you do some testing and find that it needs improvements to make it useable, I’d be willing to have a look to see what can be done. I have limited time and it’s been years since I wrote this program, so if you’re willing and patient, we should be able to make some headway on this.
I started doing some more testing. I’m trying a “newer” computer, a little netbook that I would most likely using on jobs (still XP however) and also updated the drivers for my Edirol UM-1S midi interface. I’m going to let it run most of the day and see if it is stable.
I did confirm however that the program does NOT reset the attenuator levels when you “deactivate” it. That could be quite a surprise if you decided to suddenly switch to a manual mode!
Thanks for checking that, Tom. I thought I had added a function that returned the attenuators to 0 when deactivated. I guess not! :icon_redface:
It’s something I can change, but let’s first see whether or not the rest of it works properly for you and if there’s other parts of the program that need attention.
This might be part of the problem. I apparently had an old version I was trying, version 1.1.2.0. I’ll contiuue testing with the newer 1.1.3.1 and see how it goes. I thought I had checked that before I began testing. I’m sorry.
Okay. You are correct! With version 1.1.3.1, when I click on"STOP", the program DOES reset all the attenuators to zero! HOWEVER, about 1 out of every 3-4 times that I click on “STOP”, the program locks up and does NOT reset the attenuators. By the way, this seems to be the same spot where v.1.1.2.0 was locking up, when I clicked on “STOP” (but again not all the time).
In reference to the lockup problem, check that on the LS9 you have ONLY parameter change RX & TX selected and no “Echo” or anything else selected.
It’s a problem that happens on a few MIDI interfaces due to the non-standard way they handle the “Stop Recording” message. I had to build an elaborate work-around on my SOF app to get around it.
If this program seems useful or potentially useful I will address that problem if it continues to happen.
I think I downloaded v.1.1.2.0 sometime last year. I thought I had checked the version number against the current version you have posted, but obviously I goofed!
Since my first test of the Automixer program I have ONLY had Parameter TX & RX selected. HOWEVER, the LS9 in the midi menu also has the the option in the “MODE” box of:
SINGLE / MULTI (you must select one or the other)
OMNI / BANK
NRPN / TABLE (you must select one or the other)
Right now I have selected:
MULTI
(omni & bank are both off)
NRPN
“Program Change” is all off
“Controll Change” is all off
“Other Command” is all off
The Single/Multi/Omni/Bank stuff has to do with Program Change messages (which we’re not using) and the NRPN/Table has to do with Control Change Messages (whcih we’re not using)
So, let’s see first if it works for you as an automixer and then get on to bug fixes and other enhancements.
From all I can tell from “listening” tests, its basic function as an automixer is excellent. I’ve been doing most of my experimentation with 4 lav mics (haven’t tried a full 8chs yet). It clearly gets rid of the hollowness (comb filtering) and excessive room ambiance associated with having multiple open mics in proximity with each other.
I don’t currently have a job scheduled where I need a automixer, but as soon as I do, I will give it a shot (assuming it’s not a high profile event). None the less, while the automixer is active, it appears to be VERY stable. I’ve had it running most of today with no problems. I’m going to let it run all night and see what happens.
The" lock ups" that I have been experiencing may not be actual “lock ups”. It happens when I click on" STOP", but not every time. When it does, it “seems” like the program is locked up, and will not resume when I click on “GO”, however if I first click on the “MIDI” button, then on “OK” (the midi selection remains intact and I do NOT have to reselect it - just clicking on “OK” is enough), then the automixer restarts immediately and runs just fine.
But YES! I believe your automixer is going to work VERY well! I should be getting a bunch of PCC mics delivered to me tomorrow. I use them alot on video taped “table” discussions, and as a result are very familiar with their performance using a Shure SCM810. I’m looking forward to seeing how they do with your program
You have discovered the “workaround” that’s necessary for some MIDI interfaces. The “Stop” command needs to be followed by a MIDI close command. Then to resume you have to open the ports again.
Are you finding that the attenuators are or are not being set to zero when you hit “stop” (with or without the lockup)?
The attenuators ARE resetting to zero with this type of “lock up”, HOWEVER, I just got a different type of lock up (like the ones I was getting the other day) where after clicking on “STOP” I get the Windows “hour glass” and the program is completely unresponsive. I can click on the “X” on the top right on the window to exit the program and after that I get a Windows message that to program is “not responding”. Finally the program closes and Windows gives me to option to report the problem.
Also, something new but perhaps related to the midi close command you referance above, I noticed a couple times when I clicked on"STOP", the program lost the indicator that is was controlling an LS9 and asked that I move a fader again. I had not seen this before. Moving a fader or two (sometimes it took quite a bit of fader movement to be recognised as an LS9 again) seemed to finally registar and I was able to click on “GO” and continue.
I let the AutoMixer program run all last night and most of today. It seems quite stable while in the “active” mode.
I did however discover one thing that almost always causes it to lock up. If I click on the “MIDI” button while the program is “active”, I then MUST click on the “OK” button to get out of that window and the program nearly always locks up. I assume that would be an easy fix by simply NOT allowing the user to open the MIDI window while the AutoMixer is “active”.
One additional feature I’d like to see would be a “dB per second” setting for how quickly the gain rises. I find that after someone ceases to speak, the “jump” up in gain of the other channels (room ambiance) is it a bit too obvious. Believe it or not, I’d love to be able to slow it down. I think this would sound more natural.
