ATEM Mini Audio?

StradivariusBone

Custom Title
Premium Member
Fight Leukemia
Like everyone else on the planet, I am now becoming an expert in live and not-so-live video production. The church I work for has recently acquiring an ATEM Mini switcher to eventually feed a livestream. For now we're just doing pre-recorded. Our workflow has been to record multiple cameras all feeding the switcher and capture audio from an M32 with a laptop configured with Dante Virtual Soundcard (which works great!). Since it's not live, I'll do some work on the tracks on the laptop, send the audio to my boss who then stitches it together with the video as needed. We've tried a few things to feed the ATEM mini with an audio feed however, with limited success. So far it's just distorted and sounds like it's clipping the input to the ATEM, but it's fine at the sources. There doesn't appear to be any way to monitor audio at the ATEM either.

First, it does have two inputs labeled Mic 1 and 2. From everything I've read online, it looks like these are looking for unbalanced stereo input, like a standard 3.5mm headphone connection. I configured an output from the DVS laptop to send and we got signal, but on the playback it's always distorted. Then it occurred to me yesterday that the HDMI connections are probably also feeding in audio from the cameras. This ATEM Mini seems to be marketed toward fancy YouTubers, so it would make sense that they'd utilize camera audio as well. The software that configures the switcher seems limited at best, and I haven't had a chance to really dig into it yet, but I can't seem to find anything to modify other than enabling the mic 1/2 inputs and adjusting the input gain.

My thought now is to bypass the switcher entirely for audio and feed it directly to a computer that will run OBS via some sort of USB interface. I'm not sure if OBS has the capability to compensate for latency though. Longterm thought is to use DVS running off the Dante to send audio to a control room that's isolated acoustically and then inject that with the video feed to go live. Is this a workable path?
 
Like everyone else on the planet, I am now becoming an expert in live and not-so-live video production. The church I work for has recently acquiring an ATEM Mini switcher to eventually feed a livestream. For now we're just doing pre-recorded. Our workflow has been to record multiple cameras all feeding the switcher and capture audio from an M32 with a laptop configured with Dante Virtual Soundcard (which works great!). Since it's not live, I'll do some work on the tracks on the laptop, send the audio to my boss who then stitches it together with the video as needed. We've tried a few things to feed the ATEM mini with an audio feed however, with limited success. So far it's just distorted and sounds like it's clipping the input to the ATEM, but it's fine at the sources. There doesn't appear to be any way to monitor audio at the ATEM either.

First, it does have two inputs labeled Mic 1 and 2. From everything I've read online, it looks like these are looking for unbalanced stereo input, like a standard 3.5mm headphone connection. I configured an output from the DVS laptop to send and we got signal, but on the playback it's always distorted. Then it occurred to me yesterday that the HDMI connections are probably also feeding in audio from the cameras. This ATEM Mini seems to be marketed toward fancy YouTubers, so it would make sense that they'd utilize camera audio as well. The software that configures the switcher seems limited at best, and I haven't had a chance to really dig into it yet, but I can't seem to find anything to modify other than enabling the mic 1/2 inputs and adjusting the input gain.

My thought now is to bypass the switcher entirely for audio and feed it directly to a computer that will run OBS via some sort of USB interface. I'm not sure if OBS has the capability to compensate for latency though. Longterm thought is to use DVS running off the Dante to send audio to a control room that's isolated acoustically and then inject that with the video feed to go live. Is this a workable path?
@StradivariusBone
Several things are incorrect: (I believe we've discussed this one once or twice before here on Dave's Control Booth Forum.)
a; The headphone output is MUCH too high in level for an unpadded microphone level input (akin to applying 277 /480 to a 120 volt lamp).
b; The headphone output will provide two unbalanced outputs, one left and one right: Two unbalanced outputs do not equate to one balanced input.
c; When sounds are mixed to stereo outputs, most mix engineers place bass down the centre to use your stereo woofers efficiently; like wise vocalists and featured soloists are routinely panned straight down the centre.
This works well with two unbalanced / mono inputs NOT SO MUCH with one balanced input.
With a balanced input, signals of equal levels arriving 180 degrees out of synch' cancel for a theoretical net level of ZERO normally resulting in little to no bass, extremely distant, hollow, barely there vocals (think budget engineered vocal eliminators) with only echoey stereo reverb effects remaining of anything panned down the centre.

How do you solve this / get to where you want to be?
Any / all of several ways:
1; Use your stereo headphone output to feed two, unbalanced, line level inputs.
2; Use a fancy direct box purpose designed for exactly this use.
3; Use two direct boxes with input pads to pad each of your two unbalanced signals down to balanced microphone levels then drive your two balanced microphone level inputs.
There's a few points to ponder.
Toodleoo!
Ron Hebbard
 
Last edited:
I'll let others chime in on the ATEM, but I can address your OBS questions.
You have a couple of options for latency adjustment in OBS, depending on which way you need to adjust.
The place to adjust audio is Edit>Advanced Audio Properties. That'll let you adjust delays on your audio inputs, including negative delays. There are reports that adding longer delays gets glitchy by chewing up buffer, but I've never had a problem with smaller delays.
Video is delayed by adding a filter to the source. Right click on it > Filters > Hit the "+" under "Audio/Video Filters" > Video Delay
 
A 3.5mm microphone input is likely not a balanced input, but rather intended for a computer-style condenser microphone (with "plug-in power", which is just a low biasing voltage on the order of a few volts present). A (pair of) passive direct boxes and appropriate cabling would be as good a solution as any for an analog connection. Padding may or may not be required in practice; but a lot of direct boxes do have a pad in any case. As to how the appropriate cabling would be wired, I can't say offhand; Google may be your best friend for that.

Looking at the manual, though, they do suggest you can plug a music player or similar device into the inputs as well as a microphone. That's pretty strange. Even more annoying is the utter lack of anything resembling technical data on these audio inputs, such as what input levels they accept or what pinouts they use or whether they are in fact stereo. You could try reducing the input gain in the mixer page of the control software thingy to see if that prevents clipping.

Frankly, the audio side of things really looks to me like something cobbled together by people who understood video but not professional audio.
 
I'm tempted to say use Dante into the laptop, but there's a catch. Dante really needs to be on a dedicated network port. If the stream goes out on wifi, it might work. Dante has very low latency, so I doubt that will be an issue.

If you want to get audio into into the ATEM, it would help to know what the mic jack is. ATEM offers no specs, but they show an 1/8" TRS being plugged in. Is it stereo, or is it balanced? Plug in a stereo cable, such as a TRS to dual RCA, turn up the gain, and induce a buzz into one channel at a time by touching one RCA center pin at a time. If you hear the buzz in just one channel, it is stereo.

If you can scrounge up some sacrificial audio cables and resistors, and are interested in doing some soldering, I can sketch up a simple attenuator circuit to properly match the M32 to the ATEM.
 
Looking at the manual, though, they do suggest you can plug a music player or similar device into the inputs as well as a microphone. That's pretty strange. Even more annoying is the utter lack of anything resembling technical data on these audio inputs, such as what input levels they accept or what pinouts they use or whether they are in fact stereo. You could try reducing the input gain in the mixer page of the control software thingy to see if that prevents clipping.

The documentation on the ATEM mini is sparse. It does claim that each "mic" is actually a stereo in, but I'm assuming that it's TRS L/R/Common that it's looking for. But it also says that it takes plugging in a condenser mic. I know my computer audio ports are now configurable to accept line in/headphone out/mic in/etc, so I'm guessing that might be what this does? I haven't been successful in finding the right setting for that. Complicating matters is the HDMI carrying audio from the cameras, which I hadn't considered until yesterday. Google-Fu is coming up way short.

Frankly, the audio side of things really looks to me like something cobbled together by people who understood video but not professional audio.
This is exactly what I told my boss LOL

The Dante side of things has been performing remarkably well. It took some configuring as there are a few known issues with DVS and Mac OSX right now, but once I sorted those out it has not been a problem. The DVS is running at 10ms of latency so it is noticeably different from live, but I have no clue how backed up the video feed is. I'll try the trick with a stereo cable to see what it gets. In the software it does show level indicators running on two inputs per channel. My first thought was to run it with input from DVS and a direct out from the M32 that I would mix live (we do a 2-track to USB backup recording) so with that we would in effect have 4 safety recordings for audio (USB recording, input to ATEM from M32, DVS to MacBook running StudioOne, output from DVS Macbook into ATEM).

With the DVS MacBook, I configured an aggregate device using USB interface that has a headphone monitor port out. So StudioOne sees the 32 channels coming in to the Mac and then buses the main mix out to the USB device. That works great, so I know from that point the audio being fed into the ATEM is fine. Something is going haywire in how it's being processed after that I suppose. And yeah, Google and BlackMagic documentation is not helping, which is where I reached out here thinking someone had seen it before.
 
There's also a third way to do this. If you take the Dante card out of the M32, and install any card with a USB port, you can connect to the laptop via USB. With the driver's installed, it becomes a sound device to the laptop. My worry is that the laptop is already getting video via USB, so that hardware might be drinking from a firehose.
 
There's also a third way to do this. If you take the Dante card out of the M32, and install any card with a USB port, you can connect to the laptop via USB. With the driver's installed, it becomes a sound device to the laptop. My worry is that the laptop is already getting video via USB, so that hardware might be drinking from a firehose.

We actually have two computers setup for this. The Dante laptop is only running StudioOne and Dante Virtual Soundcard. The ATEM Mini is set up with an iMac that we use for running ProPresenter. And yes, at least one of the cameras has an XLR line input, stereo no less. You're a genius, FMEng! I think that's our next trick to attempt!
 
Just jumping in without reading all the comments....
There is a full mixer built into the ATEM mini!

The gain on the input is likely too high, all you have to do is connect a computer with the ATEM control software and BOOM full access to all the controls not on the surface. Including a comprehensive mixer! I got an ATEM mini right before the $^# hit the fan and love it. Currently have a streaming setup on my desk.

EDIT:

You can get the software (and the update patch for the ATEM) on the blackmagic website. It is NOT laid out the best. you are looking for https://www.blackmagicdesign.com/support/family/atem-live-production-switchers
ATEM Switchers 8.2 Update
Install this on your control computer, you can use this to control the ATEM and to apply updates along with configuring network settings etc. Just use the USB-C connection
 
Last edited:
A 3.5mm microphone input is likely not a balanced input, but rather intended for a computer-style condenser microphone (with "plug-in power", which is just a low biasing voltage on the order of a few volts present). A (pair of) passive direct boxes and appropriate cabling would be as good a solution as any for an analog connection. Padding may or may not be required in practice; but a lot of direct boxes do have a pad in any case. As to how the appropriate cabling would be wired, I can't say offhand; Google may be your best friend for that.

Looking at the manual, though, they do suggest you can plug a music player or similar device into the inputs as well as a microphone. That's pretty strange. Even more annoying is the utter lack of anything resembling technical data on these audio inputs, such as what input levels they accept or what pinouts they use or whether they are in fact stereo. You could try reducing the input gain in the mixer page of the control software thingy to see if that prevents clipping.

Frankly, the audio side of things really looks to me like something cobbled together by people who understood video but not professional audio.

It is intended as a "professional mic input" to be used with XLR to 3.5mm adapters. Not sure how they have that working exactly... Some kind of switched circuit?
Most people I know who use ATEMs usually connect microphones to cameras, and then feed that signal to their switcher together with the video.
ONE of the important reasons why this is typically done is that there is a built-in latency (lag) on HDMI and similar connections. So if you want audio and video 100% in sync its best to connect the mic to the camera anyways.
When feeding audio outside of your switcher keep this latency in mind if you want to keep a good sync. Delay the output to match as best you can.
 
So if you want audio and video 100% in sync its best to connect the mic to the camera

So in that case we'd want to use a mixbus pair instead of DVS. I'm downloading the software on my personal computer now to see if I can figure it out. I found the page where audio is controlled, but a lot of it was greyed out and it was hard to see where everything was being routed. It looks like it all just forces to get mixed to the stereo mix. I feel like I'm missing a page where a lot more settings are based on what I've seen online.
 
It is intended as a "professional mic input" to be used with XLR to 3.5mm adapters.

That's just plain nutty. Is this how bad the industry has gotten? The device looks pretty well designed otherwise, but they couldn't find the real estate for 1/4" jacks, or better? I hope those are adapter cables because most 1/8" jacks will break from the stress of holding a heavy pair of XLRs.

The M32 can apply delay to any output.
 
It is intended as a "professional mic input" to be used with XLR to 3.5mm adapters. Not sure how they have that working exactly... Some kind of switched circuit?
If that's true, it would be really nice if the manual made some general mention of the fact, doubly so since it basically seems comprehensive and thorough and well-written.

All I could glean from it was that they were "stereo" 3.5mm jacks (which implies two channels of unbalanced audio), but that you could plug in a microphone such as a lavalier microphone (typically also unbalanced, with plug-in power, and not directly adaptable to standard XLR microphone inputs that supply phantom power) or an audio player (line-level stereo audio, I'd guess). Nothing there implied a balanced input, nor gave anything beyond a very broad hint as to the permissible range of input levels.
 
That's exactly how I interpreted it. Now I'm not a video guy, so I'm wondering if in the world of prosumer video and Youtubers that there are more devices that output something that would be friendly? My boss wanted to buy an XLR adapter but I told him without even understanding how it's wired it may make matters worse. I've been feeding it unbalanced line level stereo from a headphone monitor output and getting distortion, even though the levels in the app were set below clipping and when listening back on headphones the input mix sounded fine. Do those beltpack receivers that camera guys use have 3.5mm out? Is that what they are thinking of? I have to say, it's really well-built aside from this little headache.
 
Lookie here! Found the very inconspicuous gear icon hidden in the bottom left. We are feeding one of the mic inputs from a mixbus directly and the other from Dante. We'll see how it works in a few.
 

Attachments

  • IMG_20200416_170412.jpg
    IMG_20200416_170412.jpg
    399.1 KB · Views: 1,331
I'm an avid user of LightFactory. I am very familiar with manual lag. It's not bad now but I think at one point we were 5 or 6 versions back from where the software was.

Still recording, but so be far so good!
 
So I'm pretty confident that we're getting audio into the ATEM at an appropriate level, however the final recording was still a mess of distortion. There was more to the signal path that I didn't realize. So the ATEM mini has a USB-C that can send the output video and audio to a computer to be captured by OBS, etc. What we are doing is using the HDMI out to a HDMI to SDI converter and then into a recording device from BlackMagic called the Hyperdeck Mini. I didn't even realize that SDI could carry an audio signal, much less 16 channels worth! Anyway, I was under the impression that the computer was capturing the whole program, but I'm pretty confident the gain structure problems are coming on after it leaves the ATEM. I haven't begun to dive into the Hyperdeck manual yet, but I'm betting there's something internal there that's an issue.
 

Users who are viewing this thread

Back