Audio
The server makes every listener their own mix and dips exactly as much as you set: everything steps back under the director’s voice, and the program fades down by itself when the crew talks.
01
While someone talks, a “Dim under talk” feed dips by a set amount (24 dB by default), and under a priority talker everything else does (12 dB by default). The dip starts at once, doesn’t pump between words and comes back over half a second.
02
PGM, an AUX or any line input comes in as a feed: a computer’s browser sends it from the /feed/ page, protected by a PIN. Roles listen with a Listen key or on a fixed route; the feed page sets its input gain too.
03
Every role can have its own input gain, noise gate, speech threshold offset, master volume and release tone. The camera operator in a noisy spot gets a gate, the quiet talker a little gain – nobody else is affected.
04
The jitter buffer measures each client’s network jitter and follows it: minimal on a quiet LAN, deeper on WiFi. The bitrate is set separately for each direction, and the Engine page shows what actually flows.
Per listener: instant attack, gradual release, so even what is too loud doesn’t distort.
Opus FEC restores a single lost packet inaudibly, PLC covers several in a row.
A roger beep for the listeners when the talker lets go, at the level they heard the talker.
Your own voice in your headphones, locally, without delay.
Step by step
The server doesn’t hand out one shared channel: everyone gets exactly what they are listening to, at their own volumes and with their own ducking – and never their own voice.
Speech is clear at 32 kbps; the 64 kbps default suits program audio too. As the server sends mono, above about 128 kbps there is no audible gain, only more bandwidth (at 256 kbps, 32 clients take about 8 Mbit/s).
Yes: on a wired LAN the jitter buffer’s target depth can be 1 frame, which is 20 ms faster. On WiFi or the internet 2–3 frames and the adaptive mode are the safe choice.
Priority ducking dims everything else while a priority role (the director, say) is audibly talking. Feed ducking dims only “Dim under talk” feeds, under anyone’s speech. The two add up.
Talk on a tap or while held, listen, tally, call, Reply and Kill – on big, legible keys, on a phone too.
Roles, groups, keys, fixed routes and presets in a password-protected admin. Every change applies at once.
Contact
Questions, a demo, or a trial in your own studio? Write or call, and we’ll show you how BroadTalk fits the way you work.