VSIP

Products

One engine. Four ways to use it.

Every product below runs on the same voice pipeline — the same VAD, the same 192-dimension speaker embeddings, the same turn model. Real-time or offline, streaming or REST.

Why it matters

One voiceprint, every product.

A voice enrolled through the Speaker Profile API is the same embedding the live stream locks onto and the same one Batch Analytics attributes turns to. Enroll a customer once and every product recognises them — no re-training, no per-product setup, no separate model to keep in sync.

That is the whole architectural bet: identity is computed once, in one place, and everything else is a surface on top of it.

Start with the free tier.