Vaani Voice

The cost of a call includes telephony, speech recognition, AI reasoning, voice synthesis, the platform fee, and any enabled add-ons. Usage-based billing charges for the components you use, while fixed-rate agreements can bundle these into a per-minute rate.

If you connect a supported component with your own provider key, that component is not charged through Vaani Voice. You pay the provider directly, while other Vaani charges, including the platform fee and any components used through Vaani, still apply.

The billing pulse depends on your offering. Vaani Voice supports per-second, 30-second, and per-minute billing. It is important to confirm the configured pulse for your account as the estimator on the website does not model pulse rounding.

You add funds to your prepaid wallet, and call charges are deducted as usage is processed. Available payment methods and billing currency depend on your account. Phone-number rentals and enabled add-ons may be charged separately.

Yes, Vaani Voice offers volume agreements with negotiated flat rates or monthly usage slabs. Postpaid arrangements, invoicing, and overdraft limits are subject to approval and a commercial agreement. You can contact their team to discuss expected traffic.

WebRTC browser calls do not have a telephony charge. However, speech recognition, AI reasoning, voice synthesis, platform fees, and applicable add-ons still depend on the stack and billing arrangement you choose.

Standard models and voices are included across all plans. For premium models or voices, you should contact the Vaani Voice team to discuss availability and pricing.

Yes, the whitelabel and reseller model supports your own branding, custom rates for sub-accounts, and distributing credits from a shared master wallet. You can contact Vaani Voice to set up a commercial arrangement for your business.

Interruption stop time, how quickly agent playback stops when the caller speaks, is typically 100–200ms. Full-turn response latency, from caller finishing to first agent audio, is p50 at 600ms and p90 at 900–1200ms, affected by model choice, network, and telephony.

Yes, you can configure a custom REST tool with its input schema. The agent can collect required details, invoke your endpoint, and use the returned result in the conversation, as illustrated by the appointment booking flow.

Post-call results finish asynchronously. The transcript endpoint might return 404 while processing, and call details may show waiting messages until analysis is ready. Completed details include summary, entity, and call_eval_tag, also delivered via call_postprocessing webhooks.

Configure your webhook endpoint to receive call events. `call_ended` indicates the conversation has ended, and `call_postprocessing` delivers the completed transcript, summary, and extracted details. You should verify deliveries using the signature contract before processing them.