Audio

Program mixer with mix-minus

Per-input gain / mute, a level analyser per participant, a program bus into the PCM worklet → Matroska → ffmpeg path, and one return feed per guest that never contains their own voice.

• Audio delay slider (−500…+500 ms) for capture dongles that lag a USB mic
• Sync monitor in the stats bar: measured A/V offset, green ≤ 40 ms
• Desktop audio via BlackHole (macOS) or Stereo Mix / VB-Cable (Windows)

Soundboard, music, trimmer & ducking

Pads (one-shot / hold / loop, per-pad hotkeys, fades, solo), background music with crossfade, and ducking under your mic. The clip trimmer cuts just the bit you need.

• Trimmer: in / out, fade in / out, gain, normalize — non-destructive per pad
• Every pad is a registry action: hotkey, Stream Dock key with live state, control API / OSC

Read the doc: docs/MEDIA.md

Broadcast audio chain & loudness

Per-channel HPF → 6-band EQ → gate → compressor → de-esser → limiter as real AudioWorklet nodes, a master true-brickwall limiter and EBU R128 / BS.1770-4 metering with a delivery target.

• Aux buses with a routing matrix, generalised ducking, PFL / AFL solo, talkback to guests
• Processing + loudness on Pro; aux buses / talkback on Business
• Editor: Track + Clip mixer with automation, Essential Sound chains, auto-duck, normalize, stems

Studio Voice

An 8-stage voice chain — HPF → mains notches → spectral gate with learned room tone → expander → de-esser → 5-band EQ → compressor + glue → loudness rider → true-peak limiter — on every mic and guest, live and in the Editor.

• Learn room tone in 3 s; noise floor before / after, spectrum behind the EQ curve, in / out LUFS
• RNNoise (arnndn) + afftdn denoise in Direct mode with impulse-measured delay compensation
• Presets, hold-for-A/B, per-mic settings; the full chain on every plan (Creator+)