Livestream Audio: Fix This Before You Touch a Camera

Every organization that asks me to improve their livestream believes they have a camera problem. Nine times out of ten they have an audio problem wearing a camera problem's clothes. Viewers forgive soft focus, dated graphics, and awkward angles. They leave over audio, usually inside two minutes, and they rarely file a complaint on the way out.

I have run weekly multi-camera church livestreams for two decades, in rooms ranging from a rented school cafeteria to purpose-built sanctuaries, with crews ranging from seasoned operators to a teenager who learned the board on Thursday. Across all of it, one pattern holds: when the stream got better in a way viewers actually noticed, the improvement almost always happened at the console, not the tripod. And the painful part is that the audio fixes are usually free, while the camera fixes are usually expensive, which is exactly backwards from where the budget requests go.

There is a reason audio is unforgiving in a way video is not. Viewers experience video as information and audio as environment. A soft shot is a flaw they observe. A harsh mix is a discomfort they feel, and people do not sit still inside discomfort. They just tap over to something else, often without being able to tell you why. Your analytics will show a two-minute average watch time, someone will conclude the content needs to be punchier, and a camera will get purchased. I have watched that exact loop run in more organizations than I can count.

Here is where stream audio actually goes wrong, in the order I fix it.

You are sending the wrong mix

The most common failure is not technical. The stream is receiving the house mix, balanced for a room full of bodies and acoustics, and it sounds hollow and lopsided online. Vocals buried, drums huge, the room's energy nowhere. The fix is a dedicated stream mix: an aux send or a separate console layer balanced for headphones and phone speakers, because that is where your online audience actually lives.

Understand why this happens and it stops being mysterious. A house engineer mixes to fill in what the room is not already providing. In most rooms, drums and stage amps are plenty loud acoustically, so they barely appear in the house mix. Vocals need the most reinforcement, so they dominate the sends. Feed that same mix to the stream, and everything the room provided for free is missing, and everything the room needed help with is shouting. It is not a bad mix. It is the right mix for the wrong audience.

The dedicated stream mix flips the logic: it has to be complete, because the listener at home gets nothing from your room except what you send. Full drums, full instruments, vocals sitting on top, and, crucially, a pair of audience mics pointed at the room. Those ambience mics are the cheapest transformation available in this entire discipline. Without them, three hundred people singing sounds like one person singing into a void. With them, the listener is in the room. Two decent condensers on the back wall, faded in during congregational moments, will change the character of your stream more than any camera you will ever buy.

If your console cannot spare an aux, or your volunteer bench cannot staff a second mix position every week, there are still moves available: a matrix output with its own EQ and a heavy dose of the ambience mics gets you most of the way. But if you are choosing between a camera upgrade and a console with enough buses to build a real stream mix, buy the console. Every time.

Nobody is listening to the actual output

Every week, someone must monitor the real stream output on headphones. Not the board feed. The thing that leaves the building. Half the disasters I get called about, a dead channel, a missing mic, thirty minutes of hum, would have been caught in seconds by one person wearing headphones with the actual program in them.

The distinction between the board feed and the true output matters because failure hides in the gap. The signal leaves the console healthy, then passes through a USB interface, an encoder, a streaming computer, and a platform, and any link in that chain can fail while the console meters glow a reassuring green. I once spent a Sunday troubleshooting a stream that broadcast forty minutes of perfect video over the sound of silence. The board feed was flawless. A sample rate mismatch at the interface had killed the audio downstream, and nobody was positioned to hear it. The comments told us before the meters did, which is a sentence that should embarrass every production lead who reads it, me included.

So make monitoring a named job, not a vibe. In my rooms the rule is one person, one pair of closed-back headphones, listening to the return from the actual stream, checking on a schedule: before service, at the first song, at the transition to the message, and after any input change. Phone in hand with the public stream as a second check. It is the least glamorous role on the team and it prevents more visible failures than every other role combined. When someone asks how to get started volunteering in production, this is where I put them, because it teaches the entire signal chain one failure at a time.

Dynamics are doing nothing, or everything

Streams need gentle compression to survive the jump between a whisper and a full band, and a limiter as the final safety. But I find consoles where nothing is compressed and the quiet moments vanish, and consoles where everything is crushed and the whole service arrives fatigued. Gentle, musical, and boring is the target.

The reason dynamics matter more online than in the room comes down to where people listen. In your sanctuary, the quietest prayer and the loudest chorus both work, because the room and the human ear handle an enormous range gracefully. At home, that same range has to survive a phone speaker at a kitchen table with a dishwasher running. Uncontrolled, the listener rides their volume button all service: up for the prayer, dive for the downbeat. Every trip to the volume control is an invitation to leave.

The fix is layered and modest. Light compression on the vocal channels so words stay intelligible. A gentle bus compressor across the stream mix, doing two or three decibels of work on the loud moments, which is enough to matter and too little to hear. A limiter at the very end of the chain, set where it almost never engages, as insurance rather than an effect. If you can hear the compression working, back it off. The settings that sound impressive on a solo are the settings that arrive exhausting after seventy minutes.

And a word about loudness: streaming platforms normalize levels now. Pushing your stream hot does not make it louder on arrival; the platform just turns it back down, and you paid for the ride in crushed dynamics. Mix at a sane consistent level and let the platform do its math.

Teach the mic before you touch the board

Here is the fix that costs nothing and is invisible on every equipment list: what happens in the first six inches in front of the microphone matters more than everything after it. No console setting rescues a pastor who drifts off-axis at the emotional peak of the message, or a vocalist whose mic distance doubles between verse and chorus, or a handheld passed between three people at three different volumes.

Ten minutes of coaching, repeated kindly and often, outperforms thousands of dollars of processing. Consistent distance. Speak across the mic, not into your chest. Lavalier and headset mics placed the same way every single week, because every placement change is a new EQ problem someone has to chase in real time. On volunteer teams this is doubly true: your rotating crew cannot re-solve a new input puzzle every Sunday, but they can absolutely run a system where the sources behave consistently. Boring sources are what make volunteer mixes good.

The order matters

Mix first, monitoring second, dynamics third, and only then talk to me about cameras. Gear upgrades are fun to shop for, and that is exactly why they get purchased before the free fixes get made. Your audience cannot see your new camera if they already left because of your old mix.

There is a budget-meeting version of this principle worth spelling out. The camera upgrade is visible, photographs well, and makes everyone feel like progress happened. The stream mix is invisible, costs mostly attention, and is the actual reason people stay past minute two. When I consult on production budgets now, I ask one question before any purchase: have we exhausted the free fixes? A dedicated stream mix is a configuration. Monitoring is a person. Mic technique is a conversation. Dynamics is an afternoon of careful settings. All of it together costs less than one mid-range camera body, and all of it together is worth more than three of them.

So here is the whole discipline in one paragraph. Build a mix for the people who are not in the room. Put a human being on the real output every week. Control the dynamics gently enough that nobody notices. Coach the six inches in front of every microphone. Then, when all of that has been boring for a season, buy the camera. It will finally have an audience still watching when the picture improves.