← livestreamautomixer.com

LAM — LiveStream AutoMixer

What LAM does

LAM listens to each channel coming from your mixing board and continuously balances them for the livestream — keeping the lead vocal audible over the band, holding a steady loudness so the stream doesn't blast during the loud stretches and vanish during the quiet ones.

It does not change what the room hears. Your sound tech mixes for the room as always; LAM mixes for the people watching at home.

Mixing console
or interface
every channel, separately
LAM balances the mix as it plays
Program
Streaming software
Monitor
Your headphones

Program and Monitor carry the same audio. Monitor is simply a second copy, so you can hear exactly what the stream is getting — until you press Solo, which changes only the Monitor.

Every channel arrives separately over USB or Dante. LAM balances them and sends one finished mix to both outputs — Program to whatever software is streaming, Monitor to your headphones so you can hear what the stream is getting.
The LAM main window: channel strips across the top, master bus on the right, processing for the selected channel below
The main window. Channel strips run across the top — one per microphone or instrument — with the master bus on the right. Channel 1 is selected here, so the EQ, gate and compressor along the bottom are showing what LAM is doing to that channel. You don't adjust those; they're there so you can see the reasoning.

Setting up

1. Choose your input

Pick the device your board sends audio to — a digital mixing console over USB, a Dante interface, or any audio interface with multiple channels. LAM needs each instrument and microphone on its own channel; a single stereo mix won't work, because there's nothing to balance.

Connect the board's USB straight to the Mac. One good cable, at most one adapter, and nothing else on that port — no hub, no chain of dongles, no webcam or display sharing it. A recording program can hide a marginal USB link behind a big buffer; LAM keeps its buffer small so the stream stays in time, and the same link shows up as audio that runs, stutters and stops. If the meters keep freezing, the cable is the first thing to change.
The first time LAM opens, macOS asks for microphone access. Allow it — that permission covers the board's USB feed too. If it was declined, or the question never appeared, LAM shows no signal on any channel: open System Settings → Privacy & Security → Microphone, switch LAM on, and open LAM again.

2. Assign each channel

For each channel, set the input it comes from and give it a name you'll recognise — "Lead Vocal", "Host Mic", "Bass". The name is what LAM uses when it needs to tell you something, so name them the way they're labelled on your board.

The input picker showing all 32 channels arriving from a digital mixing console, with green dots marking the channels that currently have signal
Picking an input. Every channel your board is sending appears here — and a green dot means audio is actually arriving on it. That's the fastest way to confirm you've grabbed the right one: get someone to talk into the microphone and watch which number lights up. Far better to find a wrong assignment now than once you are live.
A channel strip showing its name, type badge, level meter, AUTO amount and fader
A channel strip. The name ("Lead Vocal") and the type badge ("LEAD") are the two things worth getting right. LEVEL is what's arriving from your board, and AUTO is how much LAM is adding or removing right now — here it's finding the vocal 12.3 dB too quiet and lifting it. The input for this channel is chosen at the bottom.

3. Set the channel type — this is the important one

The type tells LAM what each channel is, and that decides how it's treated. A lead vocal is held forward; drums sit back; a speaking microphone is levelled differently from a singing one. Getting types right matters more than any other setting.

The channel type list, with a search box at the top and types grouped by family — Vocals, Guitars and Keys, Bass, Drums and Percussion, Strings, Brass, Woodwinds, Playback — with Lead Vocal selected
The type list, grouped into Vocals, Instruments and Playback. If a channel doesn't obviously fit, pick the closest thing in the right group — the group matters more than the exact choice within it. The one to be deliberate about is Speech / Spoken Word: it's levelled for talking rather than singing, and it's kept dry.

4. Save your setup — do this before you walk away

Saving turns everything you just did into a profile. Open Profiles in the top bar, give it a name you'll recognise — "Sunday", "Midweek", "Youth" — and save.

LAM will not ask you to do this. Nothing prompts you, and nothing warns you when you quit. If you close LAM without saving, every channel name, type and input you just set is gone, and you start again from an empty desk. Save now, while you are still sitting here — not after the first service, when you have already stopped thinking about it.

Save a fresh profile whenever your setup changes — a new singer, a different instrument, a microphone moved to another channel.

Once saved, LAM reopens with your last profile already loaded — and starts audio by itself. Channels, names, types, inputs and your output devices all come back, so a normal show day is: open LAM, check the profile name in the top bar is the one you expect, and you are running. If the name isn't right, load the correct profile from Profiles.

Audio only restarts on its own if the device you send Program to is switched on and connected when LAM opens. If your interface is unplugged, or powered up after LAM, output stays stopped — open Configure, check the device is listed, and start it there. That is worth knowing before a service rather than during one.

5. Choose your output and start audio

Outputs are set behind Configure in the top bar — the top bar only displays which devices are chosen, so clicking the device names does nothing. Open Configure and set two things:

Then press start. Nothing reaches the stream until output is running — LAM can look completely normal, with meters moving, and still be sending nothing.

The top bar showing the saved profile, Program and Monitor outputs, and Configure
The top bar. Program is what goes to the stream — a Dante interface here; Monitor is what you listen on. The loaded profile is named beside Profiles ("Main Room"), so you can tell at a glance that the right setup came up. The start control sits just to the right of Configure — until audio is running, nothing reaches your stream.

Sending audio to your streaming software

If your streaming software runs on a different machine, you don't need this section — send LAM's Program output to whatever interface feeds that machine, and you're done.

If your streaming software is on the same Mac as LAM, the two can't simply be pointed at each other: macOS has no built-in way for one app's output to become another app's input. LAM can hand its mix over directly, which needs nothing installed — but not every streaming app can accept it that way, so there's a second route that works with all of them. The table below says which one you need.

Option A — straight from LAM (recommended)

LAM can hand its mix directly to your streaming software. Nothing to install, nothing to configure at the system level.

  1. In OBS, add a Media Source. (These steps use OBS as the example; the table below says what to use in other software.)
  2. Untick Local File.
  3. In Input, paste: http://127.0.0.1:3000/api/audio-stream

That's it. The address is LAM talking to itself on the same machine — nothing leaves your computer and nothing needs to be installed.

This works if your streaming software can take a web address as a source. Not all of them can:

SoftwareUse
OBS StudioOption A — Media Source, as above
Streamlabs DesktopOption A — built on OBS, identical steps
WirecastOption A — but add a Web Stream source instead of a Media Source, same address
Ecamm LiveOption B — takes microphones, devices, files and app audio, but not a web address
Meld StudioOption B — on macOS it accepts audio input devices only
StreamYardOption B — it runs in a browser, which only sees microphones
vMixWindows only, so it will be on another computer — send LAM's Program output to whatever interface feeds that machine
Line the sound up with the picture once. Every audio path has to be matched to the camera — that's true with or without LAM, on the same computer or a different one. With LAM in the chain the sound usually arrives a little after the picture; a few hundred milliseconds is normal. In OBS, open the Filters for your camera or capture-card source, add Video Delay (Async), and raise the delay until lips and sound agree. If it's the picture that arrives late instead, delay the sound: open Advanced Audio Properties (the ⋮ or gear next to the source in the Audio Mixer, or Edit → Advanced Audio Properties) and set a positive Sync Offset on the audio source. Either way you set it once; re-check it only if you change how audio reaches the computer.

Option B — a virtual audio device

If your software is in the second half of that table, this is the way. A virtual device makes LAM look like an ordinary microphone or sound device, which every streaming app understands. It's an extra piece of software to install and keep working across macOS updates — which is why it isn't our first suggestion — but it works everywhere.

  1. Install BlackHole 2ch (free) and restart the Mac.
  2. In LAM, set Program to BlackHole 2ch.
  3. Set Monitor to your headphones or speakers.
  4. In your streaming software, add an audio input source — in OBS that is Audio Input Capture — and choose BlackHole 2ch.
Don't skip step 3. Once Program is going to BlackHole, that audio goes straight to the stream and nowhere else — your Mac will be silent. Monitor is what you listen on, and it's a separate output for exactly this reason. A silent room here is normal and does not mean the stream is silent.
Whichever you choose: nothing reaches your stream until LAM's output is running. This is far and away the most common thing people call about, and it isn't a fault — the stream is fed from LAM's live output, so if audio isn't started, there is nothing to send. Start audio in LAM first, then check your streaming software.

Channel types

Four channel strips showing different type badges: LEAD, BGV1, BGV2 and AC.GT
Each strip shows its type as a badge. The badge is what LAM acts on — the name is only for you. A channel called "Acoustic" and typed KEYS would be treated as a keyboard, so it's worth a look down the row before you go live.
TypeUse it for
Lead VocalThe lead voice. Everything else is balanced around this.
BGV 1 / 2 / 3Backing vocals, in order of prominence.
SpeechSpeaking microphones — a host or speaker, announcements, a lectern or podium.
Choir (beta)A microphone over a group of singers. Treated as a backing vocal.
Acoustic / Electric Guitar, Keys/Piano, Organ (beta), Mandolin / Banjo / Ukulele (beta)Guitars and keyboards.
Bass, Double Bass (beta)The low end. Upright bass lives here too.
Drums (Full Kit), Kick, Snare, Toms, Hi-hat, Overheads, Percussion (beta), Electronic Kit (beta)Use Drums (Full Kit) for one or two mics on the whole kit or a drum mix from your board. Use the pieces when each drum has its own channel. Percussion is cajon, djembe, congas and the like.
Violin, Viola, Cello (beta)Bowed strings.
Trumpet, Trombone, French Horn, Low Brass (beta)Brass. Low Brass is tuba, euphonium and sousaphone.
Saxophone, Flute, Clarinet, Harmonica (beta)Woodwinds and reeds.
Other InstrumentAnything not listed. Sensible, general-purpose settings.
Beta means the settings follow the same standards as every other type but haven't yet been confirmed on a real instrument. They work; they may get a small adjustment once they have been heard in the field.
MultiTracksPre-recorded stems playing alongside the live band.
Background TrackMusic from a computer — countdowns, walk-in, video audio.
Room / CrowdAudience microphones. Never boosted — they add atmosphere, not content.

On the day

  1. Turn your interface or console on before you open LAM.
  2. Open LAM. Your profile loads and audio starts on its own.
  3. Check the profile name in the top bar is the one you expect, and that output is running. If it isn't, open Configure and start it.
  4. Check levels at soundcheck, not five minutes before you go live.
  5. Leave it alone. LAM adjusts continuously — it doesn't need supervising.
Do your level check at soundcheck. LAM will tell you if a channel is coming in too hot or too quiet, but it can only tell you once audio is flowing. An hour before you are live there's time to fix it calmly; ten minutes before, there isn't.

When LAM asks for your help

LAM fixes most things on its own. The one thing it can't fix is how much signal a channel is sending it in the first place — that's set at your board. When a channel is arriving so quiet that LAM has run out of lift, or so loud that it's about to distort, a strip appears at the top of the window naming the channel and saying what to do.

The alert strip at the top of the LAM window, expanded, naming a channel that is too quiet and saying how far to turn its gain up at the board
The alert strip, expanded. It names the channel by the label you gave it, says which way to move the gain and roughly how far, and counts how many channels are waiting ("1 of 2"). The arrows step between them; dismiss waves one away until it comes back for a real reason.

It comes in two levels, and the difference matters:

LAM judges this over songs, not seconds. A quiet verse doesn't count against a channel and a loud chorus doesn't count for one — it's asking about how the channel is set, not how the song is going. So the best time to see these is during rehearsal, when there's time to walk to the board. Once you're live it stays quiet unless something genuinely changes.

The one exception is clipping. If a channel actually hits the ceiling — someone cranks a playback device to full, say — LAM tells you at once, because that damage happens before LAM can help and nothing downstream can undo it.
"If nobody is using this mic, switch it off." A microphone that's left on but not being used looks, to LAM, exactly like one that's too quiet — and LAM will be turning it up as far as it can, bringing room noise into your stream. When you see that line, the fix may be the mute button rather than the gain knob.

Frequently asked questions

I'm getting no signal in my streaming software.

Almost always: LAM's output isn't running. LAM only sends audio while output is started. Start it and the signal appears.

If output is running and your streaming software still hears nothing, check how the two are connected — see Sending audio to your streaming software.

A channel is much quieter than it should be.

Usually the gain for that channel is set too low at the board. LAM can boost a quiet channel, but only so far — past that it runs out of headroom and the channel stays buried. If that's happening, LAM will tell you which channel and roughly how much to turn it up.

LAM says a channel is "too loud coming into LAM."

That channel is arriving louder than it needs to be and is close to distorting. Turn its gain down at the board by roughly the amount LAM suggests — never more than that; LAM only ever asks for enough to bring the channel back to where it should sit. It clears on its own once corrected.

Sometimes LAM will say a channel is clipping but that its level is already right, and that turning it down would make it too quiet. That's a very dynamic source — a singer with a big range, usually. Gain can't fix that; a compressor on that channel at the board can.

Turn down the gain (or trim), not the fader — LAM usually receives the signal before the fader, so moving the fader may change what the room hears without changing what LAM gets.

LAM says a channel is "too quiet coming into LAM."

LAM has turned that channel up as far as it can and still isn't getting enough. Turn its gain up at the board by about the amount it names. If the strip says nobody may be using that mic, check whether it's simply been left on — muting it is the fix in that case.

The sound crackles or breaks up.

Usually one of these:

LAM is showing a notice about my license.

Most of the time LAM says nothing about its license — a subscription renews on its own. A notice appears only when something needs you:

Any notice can be dismissed, and it only comes back if things get more urgent.

How do updates work?

LAM checks for a new version on its own and downloads it quietly in the background. It never installs anything while you're running. When you quit LAM and an update is waiting, it asks one question: install it when LAM closes? Say yes and the new version is there next time you open it; say not now and it'll ask again another day. You can also look any time from Help → Check for Updates.

Do I need to adjust LAM while I am live?

No. That's the point of it. Set it up, start it, leave it.

Does LAM need an internet connection?

Not to mix audio. It does check its license periodically, but it keeps working if the connection drops.

Does LAM send anything to SungTech?

Never your audio. LAM checks its license, and looks for updates to LAM and to our terms, now and then. When you accept our terms in LAM, when you first set it up or after an update, it sends us a record of that. And, only if you say yes, it sends a short crash report when part of it fails, so we can fix it. LAM asks once; you can change your answer any time with Help → Send Crash Reports to SungTech. A report says what failed and which versions of LAM and macOS you have, and includes the most recent part of LAM's log. It's linked to your license, so we know it's yours. Before it's sent, LAM is built to take out the names and IDs of your audio devices, your Mac's name, and your own name and email address. LAM sends reports in the background while your audio keeps running. Help → What a Crash Report Contains has the full details.

Which version am I running?

Help → LAM (build …) shows it. For support, use Help → Copy Diagnostics to Clipboard and paste that into your email.

Where are the logs?

Help → Reveal Logs in Finder. They live in your home folder under LiveStreamAutoMix/logs.

Something sounded wrong during a broadcast. What should I send?

The diagnostics (above), roughly what time it happened, and what was going on — a song, a speaker, announcements. If you have a recording of the stream, note the timestamp. Knowing when narrows it down enormously.

Getting help

Use Help → Copy Diagnostics to Clipboard and include it in your message. It identifies your exact build, which is the first thing anyone will ask.