AudioGate

Audio notifications that know when to speak

Holds TTS during meetings. Plays when you're free. Records your voice on command. Runs silently in your system tray.

Windows 10+ · No install required · Auto-updates

Your server finishes a deploy. Your CI pipeline fails. Your home automation triggers. The TTS notification fires - right in the middle of your call.

OS-level Do Not Disturb blocks everything or nothing. AudioGate holds what can wait, plays what can't, and drains the queue the moment your meeting ends.

The Decision Matrix

AudioGate monitors your mic, audio output, and focus state. Each incoming notification is played, held, or ducked based on priority and context.

Mic	Audio	DND	Normal	High	Urgent
off	off	off	play	play	play
off	on	off	hold	play	play
on	any	off	hold	hold	play
any	any	on	hold	hold	play

Held notifications drain automatically when conditions clear. Nothing gets lost.

Two tools, one tray icon

Audio Gate

Meeting Detection

Monitors mic and audio output via Windows Core Audio. When your mic goes hot, incoming notifications hold automatically.

Volume Ducking

When other audio is playing, AudioGate lowers system volume, speaks the notification, then restores it. No more choosing between music and alerts.

Replay Buffer

Missed something? The last 20 played messages are kept in memory. Replay from the dashboard, tray menu, or CLI.

Three Policies

Auto reads context and decides. Always plays immediately. Manual holds everything until you flush.

Voice Recording

Keyword Activation

Say "open gate" and recording starts. Silence for 4 seconds ends it. No buttons, no clicks.

On-Device ASR

Keyword detection runs locally using Moonshine Tiny - a quantized ONNX model. No cloud, no API keys, no audio leaves your machine.

Silence Trimming

Silero VAD trims leading and trailing silence from recordings automatically. Clean audio from the start.

Live Dashboard

Real-time waveform, RMS meter, and ASR transcript in the browser. Play, download, or delete recordings from the web UI at localhost:9998.

Architecture

  Remote machine                       Windows client

  tts-play ──── TCP :9999 ────►  Gate ──► Player ──► Speakers
                                    │
                               Detector (polls 1s)
                                    │
                            ┌───────┼───────┐
                            │       │       │
                         Mic    Audio   DND

  Microphone ──► Recorder ──► ASR ──► Keyword? ──► Record
                                                   │
                                            Dashboard :9998

Send audio from anywhere

Pipe TTS output to AudioGate over TCP. Any machine on your network can send notifications.

From a Linux server

tts-play "Deploy complete - production v2.4.1"

HTTP control

# Check status
curl http://localhost:9998/status

# Change policy
curl -X POST http://localhost:9998/policy -d '{"policy":"manual"}'

# Flush queue
curl -X POST http://localhost:9998/flush

# Skip current
curl -X POST http://localhost:9998/skip

CLI subcommands

audiogate status          # gate state
audiogate flush           # play all held
audiogate skip            # skip current
audiogate pause           # toggle pause
audiogate replay          # list recent
audiogate replay 0        # replay latest
audiogate update          # check + apply

Entirely local

No cloud. No account. No telemetry. ASR runs on-device via ONNX Runtime. Audio never leaves your machine. Recordings stay on your disk. The only network call is checking for updates from your own staging server.

Simple pricing

One plan. Everything included. Cancel anytime.

AudioGate Starter

Everything you need

$19 /mo

Context-aware audio gating
Keyword-triggered voice recording
On-device ASR - no cloud, no data leaves your machine
Web dashboard, system tray, and CLI
Auto-updates with every release
Replay buffer - last 20 messages on demand

Billed monthly via Stripe. Cancel anytime.