Weird Stereo Audio Phasing incident

[Mike Watson]

Hi,

I had a client send me some video last week that he'd shot at a press conference with a DSLR. The audio is 2-channel and stereo - the DSLR (I'm not sure what kind) must have had a left and a right mic. When you hear the interviews, they definitely have that "stereo" sound to them.

YouTube (Short URL)

Client calls me after this video has been sent out to a mailing list telling me that people can't hear the interview sound. I play it and it works fine. Client says that it works fine on computer but not on phone. I play it on my phone and it works fine. My phone has two speakers, and when I play it, it still has that "stereo" sound. I track down a phone with one speaker and sure enough, silence during the interviews. Everything else sounds fine. My gut tells me that the problem is with the stereo thing, and it's already been edited, so rather than go back in the edit and remove one channel, I just mix the whole thing down from stereo to mono in Streamclip. Play it back and there is NO SOUND during the interviews. Open it back up in the editor and kill one channel (and pan the other to center) and it works fine. Upload the new video:

91 Seconds on the 91 January 2015 Main Street Squeeze - YouTube

and it works fine in both mono and stereo.

I understand phasing and how it works. Does this mean that the camera has a hardware problem where one of the mics is hooked up out of phase? I assume it's a built-in mic.

Any theories?

[Oren Arieli]

Sounds like the left channel was a close-mic'd source, while the right channel was ambient or on-camera mic. Depending on the phone, it might only play one channel and not the other. I'm not sure this was a case of phase cancellation, as the sound would have to be the same on L and R, but reversed in phase by 180 degrees.

[Greg Miller]

Nothing mysterious about it. The interview section has two channels with exactly the same content (both speech and ambient sound) but the phase is opposite. When played in mono, you would hear absolutely nothing.

The only exception to "nothing" are two blasts of noise (road noise?) at approximately 24 seconds and 43 seconds. These are in phase. My guess is that they were added during mixdown, to cover some edit points in the actuality sound.

Most likely the cause of this was using a single balanced mic with an XLR connector, plugged into the wrong kind of XLR-to-mini adapter, thence plugged into a camera with a 3.5mm unbalanced stereo input. Since the mic is balanced, the pins 2 and 3 would have opposite phase. If those were connected to tip and ring of an unbalanced stereo input, the result would be a recording with two channels having identical signal but opposite phase. Sum those two channels to mono (i.e. on a mono phone, or in your editing software) and the result is silence.

It should never have made it out the door. When you listen in stereo, it's immediately obvious that it's out of phase!!!

[Brian P. Reynolds]

A VERY common problem when people are using 'balanced' microphones with 'unbalanced' inputs of cameras like DSLR's. (lots of info on many audio forums)
Get the proper connection cable and the problem will easily resolved.

[Jay Massengill]

The original version of the BeachTek DXA-6 XLR Adapter from long ago had a polarity reversal error in its circuit board when used in stereo mode.
So, sometimes even when using the "correct" adapter you can run into problems.

Certainly, as mentioned there are plenty of incorrect adapters being used out there.

It always pays to check the editing results in mono before sending out the finished product.

Or if it's a long project and time is short, zoom into the waveforms at each section of the project. You'll clearly see the audio is identical but out of phase.

[Greg Miller]

Just convert it from L/R to M/S. If you see some dead spots in the left (now Main) channel, you know the original channels were out of phase. (But to reiterate, listening in stereo it's very obvious when the phase is reversed.)

[Rick Reineke]

I would surmise the same, balanced hot and cold to left/right channels, 180 degrees out of phase, when summed = total cancellation.

[Richard Crowley]

If this isn't the #1 most frequently asked question, it is certainly in the top 3. Perhaps this forum needs an FAQ list.

[Mike Watson]

Top 3 most frequently asked questions? I've worked in broadcast for 20 years, and obviously understood the theory behind this, but had never seen it happen before.

Thanks, all, for the answers! Quizzical!

[Greg Miller]

Let's refer back to the original post.

Originally Posted by Mike Watson

When you hear the interviews, they definitely have that "stereo" sound to them.

They do not. Stereo would have a spread of sounds: some coming from far left, some from far right, some from center, and some from locations all across the sound stage. The interview segment did *not* sound like that; because it was *not* stereo.

On the contrary, the interviews in your clip definitely have an "out of phase" sound to them, which is obvious within five seconds after the interview segment begins. Hearing that (or watching the mix on a vector display) leaves no doubt about the cause of the problem.

And if one obviously understands the cause of the problem, and then has a mono mix that goes silent in places, there should be no question about what went wrong.

I don't know if we've really kept a tally of the most frequent questions here, but this question has definitely come up often. Luckily, the answer, and solution, are very simple compared with some of the more esoteric questions that have come up.

Hopefully you'll catch it the next time before it's too late. Good luck!

[Richard Crowley]

Originally Posted by Mike Watson

I've worked in broadcast for 20 years, and obviously understood the theory behind this, but had never seen it happen before.

Broadcast professionals typically don't encounter this particular problem because stereo TRS connectors are uncommon. And they can typically hear the symptoms if they are monitoring properly.

On other video forums (with higher traffic) this is quite a common problem among amateurs who are new to audio. Fortunately, it is pretty easy to fix.

[Mike Watson]

Originally Posted by Greg Miller

Stereo would have a spread of sounds: some coming from far left, some from far right, some from center, and some from locations all across the sound stage. The interview segment did *not* sound like that; because it was *not* stereo.

The stereo I've heard from consumer grade cameras has always originated from two microphones that were approximately an inch apart. In my opinion, it sounds largely the same as what was heard here, although, as noted, apparently I've never heard out-of-phase sound before. With + to + and - to -, apparently I've avoided this my whole career to this point!

[Greg Miller]

Out of phase stereo (recorded with two mics, one having polarity reversed) can sometimes be a little difficult to spot, because there is still a spread across the sound stage. But out-of-phase mono (what you have here) sounds so distinctively different from anything else that it's hard to miss.

Some people describe it as coming from beyond the speakers on either side. Some describe it as coming from "inside your head." No matter what you call it, it's very distinctive.

I suggest you make yourself an "ear training" file. Intercut about 10 seconds of this interview footage, with the same length of real stereo. Use multiple samples of each, alternating from one to the other. Then play it back and listen. You will soon learn to recognize the difference. It's even more striking if you listen on cans, rather than on speakers.

Once you've heard it and identify it, I'll bet you'll catch it immediately if you ever hear it again.

[Mike Watson]

Good idea. Thanks, Greg.

[Don Palomaki]

With the demise of VHS VCRs and their (mostly) mono linear tracks the problem comes up less often. Just about everything is stereo these days.

The stereo amplifiers of the old days often had a phase reverse switch on one channel to deal with just that issue in speaker wiring (and source material perhaps).