Honest numbers
Anything we haven't measured yet is marked as an example, on every page.
Company
VoxVain builds the voice layer for developers: speech, listening, turn-taking and agents that run on our own infrastructure, with honest numbers and clear rules about AI disclosure.
Principles
Anything we haven't measured yet is marked as an example, on every page.
VoxCore runs on our servers, so latency, cost and data handling are ours to control.
Our real-time voice platform: 21 voices in 12 languages, VoxCore TTS first audio, median 0.16 s (10 Nov 2026, 129 calls, US East, production phone path). Speech rendered more than 100× faster than real time on a GPU.
Anyone can sign up for a free key. Products still in Preview are marked so on every page.
Trust and AI disclosure
“You're speaking with an AI assistant. How can I help?”
Agents answer truthfully when asked whether they are an AI.
Custom voices need the speaker's recorded consent, and consent can be withdrawn.
Agents are built to say what they don't know, not to guess prices, times or status.
Agents refuse card numbers by default. Health data needs a signed agreement first, which isn't available yet.
Contact sales
Anyone can get a free API key and start building. For VoxVain Box (on-premises) or VoxVain Shield (privacy-focused tier, coming soon), email support@voxvain.com.
Get a free API key in a minute, no card needed. Then follow the quickstart.