NO CLOUD • INSTANT RESPONSE • PRIVATE

Voice AI that runs on device.

Add voice features to your product — wake words, voice commands, and more — running on device. Launch in hours.

Get started → Explore products →
Published models free for commercial use · custom models tuned to your spec
Built for
iOS Mac Android Linux Microcontrollers Automotive Wearables Smart TV Embedded & IoT iOS Mac Android Linux Microcontrollers Automotive Wearables Smart TV Embedded & IoT
Why us

Three things we do differently.

Light, fast & accurate.

Production binaries <1 megabyte. Predictable latency. Easy on the battery.

Private by design.

Audio never leaves the user's device. Models are encrypted at rest. Works offline by default.

Trained to your domain.

Contexts set to your spec. As a result, VoxRT models understand what your users mean.

Products

Lightweight audio tools that run on-device.

Simple integration, lightning fast and private by design – with no API costs.

Custom Wake Word

Available

Wake your app with a custom phrase like "Hey YourBrand" — reliable across noise, distance, and accents.

# your wake word wake_word: phrase: "Hey Vox" threshold: 0.9 # at runtime "...hey vox, what's next" → { detected: true, confidence: 0.97 }
Learn more about wake word →

Voice Activity Detection

Available

Instantly know when someone is speaking — the building block that keeps everything else fast and battery-friendly.

Learn more about VAD →

Speech-to-Intent · end-to-end

Available

Turn what users say straight into actions your app understands — no transcript step in between.

Learn more about speech-to-intent →

Custom Keyword Spotting

Available

Detect a fixed vocabulary of your custom voice commands.

Learn more about keyword spotting →

Streaming & batch ASR

Available

Turn speech into text in real time, right on the device.

Learn more about on-device ASR →

Text-to-Speech

Planned · v2

Give your app a natural voice that speaks responses out loud, fully offline.

Multilingual training

Post v1

Training pipeline built for multilingual — English today; Spanish, French, German, Italian, Portuguese and more to follow.

Devices

Everywhere voice runs.

Same model artifacts, same Rust runtime.

Mobile

  • iOS 16+ · iPhone & iPad
  • Android 8.0+ (API 26)
Available v1

Apple ecosystem

  • macOS · Apple Silicon & Intel
  • tvOS · watchOS · visionOS
Coming v2

Desktop

  • Windows · x64 & ARM64
  • Linux · x64 & ARM64
  • Web · WebAssembly
Coming v2

Embedded & IoT

  • Raspberry Pi 3 / 4 / 5
  • NVIDIA Jetson
  • ARM SoC boards · kiosks
Coming v2

Microcontrollers

  • ARM Cortex-M4 / M7
  • Cortex-M33 / M55 / M85
  • no_std-compatible
Coming v2

Automotive

  • Android Automotive
  • Automotive Linux (AGL)
  • QNX
Coming v2

Wearables & TV

  • Wear OS · watchOS
  • Android TV / Google TV
  • tvOS
Coming v2

Server / hybrid

  • Linux x64 / ARM64
  • Same artifact, server-side
Coming v2

Most voice AI providers make you compromise. We don't.

  • ✓ Tiny footprint — runtime under 1 MB; wake-word model ~100 KB.
  • ✓ No cloud round-trip. No per-detection fees.
  • ✓ Published models free for commercial use — custom models tuned to your spec.
  • ✓ Set to your spec for accuracy.
Work with us

Onboarding partners now.

Tell us what you want to build and which devices it has to run on. Or, check out our open source models.

Get started →

Published models are free for commercial use.

Ship voice features your users can rely on.

No privacy compromises, no network dependency, no per-detection costs.

Get started →
Published models free for commercial use · custom models are paid engagements