VoxEdge
Real-time voice agents — sub-300 ms duplex, 50+ languages, on-device or cloud.
What it is
VoxEdge is a On-device Voice product from Binary AI Labs. Real-time voice agents — sub-300 ms duplex, 50+ languages, on-device or cloud.
Specs
- Category
- On-device Voice
- Version
- Current release — versioned per engagement
- Deployment
- On-device; no audio leaves the hardware
- Integrations
- Embedded C/C++ SDK, ZephyrOS
- Stack
- LiveKit · Cartesia · Vapi · Liquid AI
- Pricing
- Per engagement — contact hello@binarylabz.com for a quote
- Support
- Named engineer, business-hours SLA; 24/7 by arrangement
- Compliance
- Audio never leaves the device — the privacy story is structural, not contractual
- Streaming ASR and TTS
- 50+ languages
- Tool-use inside a duplex turn
- On-device or cloud routing
VoxEdge vs. building it yourself
Model sizing against the accelerator's memory budget is the hard part, and it is hardware-specific work.
Questions
Does VoxEdge run on-prem?
On-device; no audio leaves the hardware. Where the deployment target constrains the architecture, that is settled during the discovery sprint rather than after.
How is VoxEdge priced?
Per engagement rather than per seat, scoped from a discovery sprint. There is no public rate card because the variance between deployments is too wide for a headline number to be useful. Contact hello@binarylabz.com for a quote.
Should we build this ourselves instead?
Model sizing against the accelerator's memory budget is the hard part, and it is hardware-specific work.