Network audio
InterStream
Sends a track or a mix out on the network as NDI® and SRT, straight from the insert, while the track plays on untouched. Feed the stream, InterDirector, or a room down the hall. Switch an instance to RECEIVE and an empty track plays an NDI® or SRT feed from the network instead.
- Latency
- 0 samples
- CPU
- 0.001% on the audio thread, sending, Apple M4 Pro
- Formats
- VST3, Audio Unit
- Platforms
- macOS, Windows
- Channels
- Mono, stereo
On the panel
NDI
On by default: the track goes out as an NDI® audio feed the moment InterStream is inserted.
Receivers
How many receivers are taking the NDI feed. Zero just means nobody has picked it up yet.
Dropouts
Audio that could not go out in time, counted and held until you reset it. The track never waits.
Feed name
The track's name, and it follows a rename. A name already sent from this computer is numbered.
SRT Mode
CALLER calls a receiver at HOST and PORT; LISTENER waits on PORT for one receiver to call in, and LISTENING ON shows the address to pass on.
SRT Codec
AAC from the computer's own encoder, 96 to 192 kbps, or uncompressed 24-bit PCM.
SRT Address
HOST and PORT, where a caller sends; LATENCY, how long SRT waits to resend lost packets; STREAM ID; and a PASSPHRASE for AES-128 encryption. Paste a whole srt:// address into HOST to fill them in.
Level and buffer
Each output's own level and buffer, with a meter of what it sends. None of them touches the track.
SEND and RECEIVE
Which way this instance goes: the track out to the network, or a feed from the network onto the track.
Sending never changes the track
Insert InterStream on a track or a bus and what that track plays goes out on the network, mono or stereo as the track carries it. Every control acts on the feeds only: the track passes through bit for bit, with zero latency, whatever the outputs do.
The network never holds up the host. Sending runs on its own thread, so a bad connection never glitches the track. It shows up as dropouts on the panel instead. When the host stops, the feeds keep sending silence so receivers stay connected, and an offline render never reaches the network.
- Latency
- 0 samples, always
- Level
- -60 to +12 dB per output, off at the bottom
- Buffer
- 10 to 1000 ms per output, 40 ms by default
- A gap
- Sent as silence, never a repeat
NDI®: a feed named after the track
The track goes out as a full-bandwidth NDI audio feed at the session's own rate, named after the track, and a rename renames the feed. A second feed of a name already sent from the same computer goes out numbered, Lead Vox 2, and the panel says so: give your tracks their own names and each feed keeps its own.
The panel counts the receivers taking the feed. Nothing ever sends NDI HX.
- Audio
- Full bandwidth, the session's rate and channels
- Name
- The track's, numbered when taken
- Default
- On
SRT: a stream to one receiver
SRT sends an MPEG-TS stream, as a caller to a receiver's host and port, calling again until it answers, or as a listener that waits for one receiver to call in. It carries AAC from the computer's own encoder, or uncompressed 24-bit PCM. PCM always goes out at 48 kHz. AAC goes out at the session's rate when that is 44.1 or 48 kHz, and at 48 kHz from any other, resampled.
Paste a whole srt:// address into HOST and the port, stream ID, passphrase and latency it carries are filled in. A listener shows this computer's address under LISTENING ON, and Copy gives the whole srt:// address to hand to the other end. The passphrase and stream ID are saved with the session, never in a preset.
- Mode
- Caller or listener, port 9000 by default
- AAC
- 96, 128, 160 or 192 kbps; AUTO is 192 stereo, 96 mono
- PCM
- 24-bit, SMPTE 302M, about 2.3 Mbit/s for stereo
- Latency
- 20 to 8000 ms, 200 ms by default
- Encryption
- AES-128 with a passphrase
- Default
- Off
RECEIVE: a feed on a track
Switch an instance to RECEIVE and the plug-in becomes the source. Put it on an empty track and the track plays an NDI® or SRT feed from elsewhere on the network, in place of the track's own input. Each instance does one or the other, and each way keeps its own settings.
For NDI®, pick from the list of what this computer can see. The source is remembered by name, and if it goes away InterStream waits and reconnects when it returns. For SRT, call a sender or listen for one, from InterStream or any other sender, in AAC or PCM. A picture in the stream is ignored, and AUDIO TRACK picks which audio track to play. As a listener, Copy gives you the whole srt:// address to hand to the other end, with the stream ID and passphrase when they are set, so take care where you paste it.
A small, inaudible rate correction keeps up with the sender's clock, and DRIFT shows how far apart the two clocks are. A lost feed fades rather than clicks and comes back on its own, and DROPOUTS counts every gap you would have heard. The host is told zero latency, as it is when sending, so nothing else in the session is delayed to match: what you hear is later by BUFFER and the network.
- Latency
- 0 samples told to the host
- Buffer
- 10 to 1000 ms, 80 ms by default
- Rate correction
- At most 0.1 %, under 2 cents
- Mono track
- A stereo feed is folded to mono
And also
The first time on the network
macOS asks to let your host use the local network, and as a listener the firewall may ask to accept connections, on the Mac or the PC. Allow them, or no feed reaches anyone.
ACTIVE stops both feeds
Off, the feeds stop and receivers lose the source; back on, they reconnect. The track's own audio passes as it always does. In RECEIVE, off stops the receiver and the track's own input passes.
Specs
- Latency
- 0 samples
- CPU
- 0.001% on the audio thread of a 14-core Apple M4 Pro, sending, both outputs on, stereo, 48 kHz. Sending runs on InterStream's own thread, which this figure leaves out. RECEIVE has not been measured yet.
- Sends and receives
- NDI® audio, SRT (MPEG-TS)
- Formats
- VST3, Audio Unit
- macOS
- Apple silicon and Intel
- Windows
- x64
- Channels
- Mono, stereo
- Version
- 0.6.1
Download
Your download has started. If nothing appears, use the buttons above.
New in 0.5.1
InterStream now receives as well as sends. A new SEND | RECEIVE switch in the header turns an instance into a source: put it on an empty track, and the track plays an NDI® or SRT feed from elsewhere on the network, at zero added latency. Each instance does one or the other, and sessions from earlier versions open in SEND, as they were.
- NDI®: pick a source from the list of what this computer can see. It is remembered by name, and if it goes away InterStream waits and reconnects when it returns.
- SRT: as a caller or a listener, in AAC or PCM (SMPTE 302M), from InterStream or any other sender. A picture in the stream is ignored, and AUDIO TRACK picks which audio track to play.
- It keeps up with the sender's clock. A small, inaudible rate correction holds the buffer steady, and DRIFT shows how far apart the two clocks are.
- A lost feed fades rather than clicks, and comes back on its own. DROPOUTS counts every gap you would have heard.
- On the panel: BUFFER sets how much is held against the network's jitter, 80 ms by default, and FORMAT says what is arriving. A mono track gets a stereo feed folded to mono.
- An SRT listener now tells you its address. LISTENING ON shows this computer's address and the port, and Copy gives you the whole srt:// address to hand to the other end, with the stream ID and passphrase when they are set. It works the same in RECEIVE and in SEND. On a computer with several network addresses, the likeliest comes first, and you can pick another from the list. In another InterStream, paste it into HOST and the rest fills in. The copied address holds your passphrase when one is set, so take care where you paste it.
- A crash on Windows is fixed. After SRT had been used, InterStream could crash as it unloaded, usually as a host closed. This affected sending too, in every earlier version with SRT.
- Nothing about sending has changed otherwise: your streams, projects and presets carry on as before.
The rest of the family
Mic preamp
InterMic
A warm mic preamp for live vocals, with air, a gain rider, stage bleed taken out by a custom-trained AI model, and feedback caught as it starts to ring.
Adaptive tone
InterTone
Set the tone a source should have. It stays there as the playing changes, on vocals, guitars or drums.
Compressor
InterComp
A fast FET stage to grab the peaks, then a smooth optical leveller, with auto gain to keep the level matched.
Resonance, width, limiter
InterSilica
Shave the harshness, then seal the ceiling: a resonance suppressor, a one-knob stereo width and a limiter.
Mix balance
InterBalance
Checks your mix against a target in loudness and tone, and nudges it there slowly when you let it.
Master glue
InterMaster
One knob that glues a whole mix together and makes it sound fuller, then a limiter seals the ceiling.
Tape delay
InterTape-delay
A tape delay for vocals and instruments: each repeat comes back darker and further away, and the dry stays on time.
Reverb
InterVerb
A reverb for vocals and instruments: plates, halls, rooms and Fresh air, at zero latency and light on the CPU.
The Suite
Every Interwebbi plug-in, with one licence and one installer.