Strange problems recording with 3 USB Snowball microphones

[inkswamp]

So some buddies and I started recording a podcast a few months ago and we're having a blast doing it, but we seem to be fighting some inexplicable audio issues. We're using 3 USB microphones (Blue Snowball) running through Virtual Audio Cable and into Audacity. I'm wondering if, by any chance, any of you have a similar setup and know what might be causing our problems.

Occasionally, the audio will stutter. Other times, one or more of the mics will produce an echo. We've ruled out mic placement as a cause of the echo by the fact that the problem sometimes comes and goes mid-recording even when the mics haven't been moved. Additionally, we've had instances where the echo itself is louder than the original audio.

This is completely random, not affecting any particular one of the mics, but any of the 3 (so it's not likely faulty microphones.) We've reproduced these problems on three different laptops (one running Windows 8, the other two running Windows 7) and with 3 different USB hubs as well as plugging the microphones directly into the laptop's USB ports and removing the hub.

At this point, it's either VAC, Audacity or Windows audio drivers. I've posted around in a few other forums and have Googled this to death to no avail. Just hoping someone here might have some idea what's causing this.

[kozikowski]

I'm stunned you got three microphones to work at all.

Can I have your autograph?

USB is a time-shared system. It has to constantly reverse direction and manage itself. Nobody will notice if a keyboard or a mouse or a pallet stutters a little bit and USB disk drives can generally take all the time they want.

You're talking about three real-time, high speed, one direction data connections all fighting for priority. The instant the computer needs to pay attention to something else for a split second, it has to drop one of the balls it's juggling.

I personally would have used three different recorders. They dont have to be million-dollar laptops. You're going to get timing differences anyway, so managing sound files should not be a big deal. Did you notice yet that even though you start the show in perfect sync, you don't end that way? Yes, that's normal with three different digital devices, too.

I doubt there's anything you can do in Audacity. Audacity thinks it's recording one really complicated microphone. It will not record multiple different devices. You probably need to resolve the USB/priority/speed conflict somehow. Good luck. Snowballs are pretty old and I doubt they support any of the faster USB protocols.

That's probably why you didn't find anything on a Google search. I bet you got a lot of amazed looks.

"Wait, you're doing what???"

Koz

[inkswamp]

Thank you! I can't tell you how much I appreciate that you took the time to explain this. We had suspected at one point that it might be the result of some kind of innate issue with USB but it just didn't seem likely.

And yeah, amazingly enough, we got through a number of podcasts with 3 microphones with only minor issues. But lately, it seems we've not been so lucky.

We've discussed buying a small mixer and using 3 traditional mics so clearly that's the route to go.

Thanks again for helping me out.

[kozikowski]

But lately, it seems we've not been so lucky.

As your computer fills up, it slows down. You may be able to off-load some of your work onto another drive, inspect, error check, and defragment this one. Remember, Audacity is pounding away in there, too.

Can I assume Windows? You didn't actually say...

Do you keep a gazillion things open on your desktop and processes napping instead of closed? All slow the computer down.

Run your hardware choices past us if you want and we'll poke holes in it.

Do you have trouble with crosstalk? "[email protected]#$%. I can't make Fred any quieter because he's so loud he appears on more than one microphone."

What's your podcast address?

Koz

[flynwill]

I'm a bit surprised that this didn't "just work". Assuming that the microphones, bus, hub, etc are all USB2, the bus has enough bandwidth for about 400 microphone sound streams at 44.1 kHz 2 bytes per sample. And while everything Koz says about the computer needing to juggle all of the balls it true, with that much headroom it shouldn't be a big problem for it to do so.

If there are any USB1 devices on the bus (eg a keyboard or mouse) that could consume a big chunk of the available bandwidth (or correctly tie up the bus for big chunks of time passing relatively small amounts of data). So I would make sure that the USB bus in use had only the microphones. If this is a Windows machine you can use the "device manager" to see what devices are on each "Host Controller" to see if your microphones are sharing the bus with anything else. (This is important as the USB ports on your laptop often are on the same bus internally). If you're running OSX or Linux then there are ways to get this same information, but I'm afraid I don't know them offhand.

[kozikowski]

Yes, that is assuming nothing else going on. As in the post, it's getting worse, so the machine is choking somewhere. And that still leaves the timing problem of not all the microphones finishing the horserace at the same time. That can be deadly if a voice appears on more than one microphone.
Koz

[inkswamp]

But lately, it seems we've not been so lucky.

As your computer fills up, it slows down. You may be able to off-load some of your work onto another drive, inspect, error check, and defragment this one. Remember, Audacity is pounding away in there, too.

Can I assume Windows? You didn't actually say...

Do you keep a gazillion things open on your desktop and processes napping instead of closed? All slow the computer down.

Run your hardware choices past us if you want and we'll poke holes in it.

Do you have trouble with crosstalk? "[email protected]#$%. I can't make Fred any quieter because he's so loud he appears on more than one microphone."

What's your podcast address?

Koz

We got the same results on a brand new laptop, fresh out of the box, so it's not hard drive space.

Yes, it's Windows. We've tested both 7 and 8. Got the same results on both.

The best machine we've tested on is an HP ProBook 640 with a 2.9Ghz i7 processor and 8 GB of RAM. Not the best machine in town but more than enough power to record a podcast.

No other apps running when we record, just 3 instances of VAC's line-in/line-out windows and Audacity.

No issues with crosstalk. We've ruled out mic placement. We've gotten the same results from the mics being several feet apart and when people are talking very quietly. Additionally, we've had the problem come and go during the same sessions when the mics haven't been moved.

The podcast is hosted here: https://soundcloud.com/beer-3podcast.

If you listen to episode 12 ("The Race Card") you'll hear the echo very mildly at first (most noticeable around the 3 minute mark), but fast-forward to the 50:00 mark and you'll hear how it has built in intensity (though only on 2 microphones--one was unaffected) and has rendered the podcast almost unlistenable. Then fast-forward to 1:01:00. By that point, the echoes have suddenly disappeared and instead there's an annoying static sound coming through whenever anyone speaks.

[kozikowski]

We've gotten the same results from the mics being several feet apart and when people are talking very quietly.

Sorry. My bad. The crosstalk and delay thing is what's going to happen to you after you cure the static and breakup problem. The fact that you're using three Snowballs may mean you won't get the delay problem until your shows go to an hour or more in length, but it is going to happen.

Each microphone is generating digital sound from its own internal electronics and they're not perfect. That's what the 44100 sample rate means. All three of them are not going to exactly hit 44100. Some are going to be slightly slower and some slighter faster. They may all three be off. When you play back the show, it will be presented as if they all perfectly matched.

Your microphone may complete the hour show in 3600 seconds, but mine took 3604 seconds, putting the two shows four seconds off. Or worse.

The grownups send a synchronizing signal to each microphone, or they use analog microphones and mixers which don't have this problem.

Koz

[kozikowski]

I just now got to the sound clip — or I guess it's the actual show.

50:00 minutes.

Yes, that's totally what speed differences sound like. Because y'all are grouped around the microphones, your neighbor's microphone is picking up a little bit of your voice. As long as all the microphones are close to each other in processing speed, nobody can tell. After long enough (remember I said earlier it could take an hour or so?) the tiny speed differences mean your neighbor's microphone thinks you're speaking a split second early or late.

I can't think of a good analogy for this, but it's normal. If you picked two microphones from different makers, the effect could get a lot worse.

OK, here's a Hollywood analogy. Back when we were using movie film, the motors would push the film through the camera at the same speed the film was going to be projected in the theater. So far so good. There is an effect called "overcranking" where you speed up the camera motor a little. When you project that film in a movie theater, the picture ...appears...slowed...down.

People think when they select a sample rate for a microphone, it's perfect 44100.000. It's not. For $70 USD, they probably got the parts from a big plastic bin in a parts room somewhere and be happy if it hits 44101 to 44099.

This is where you start shopping for real microphones and a mixer.

Koz

[kozikowski]

There is one out. Remove two of the Snowballs and that fancy software thing. Set the remaining one to omnidirectional (position 3) and put it in the middle of the table. Tell everybody to lean in and get cozy.

Koz