Zendesk Builds Voice Translation Into Contact Center Calls

No Jitter reports that Zendesk is building real-time voice translation directly into its Contact Center product, turning a capability often handled through separate apps or integrations into a native call-center workflow.

The feature, announced Thursday, supports bidirectional translation on inbound and outbound voice calls.

The system lets both sides use their own languages on the same voice call, keeping the interaction in the voice channel and away from interpreter workflows or add-on desktop software.

That design choice is the center of the deployment.

A customer's language can be detected in the contact flow or selected by the agent, and the agent can control translation while remaining inside the call.

Administrators can enable the feature by queue and decide whether recordings retain original audio, translated audio or both.

Zendesk plans a closed early access program in October for eligible Contact Center customers, followed by general availability in the first quarter of 2027.

The company will not charge customers during the early access period, though use will be capped while it studies demand and operating patterns.

The first version will support Chinese, English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Russian, Spanish and Vietnamese.

Zendesk expects the list to expand after early access according to customer requirements.

The early access release will also arrive with limits that matter for contact-center operations.

Multiparty calls, transferred conversations, supervisor listen-in or barge-in sessions, and video calls sit outside the translation scope at launch.

Usage reporting and administrative usage limits will not be visible in that version either.

Pricing remains one of the open items before general availability.

Lilith Shoemaker, product marketing manager at Zendesk, described the eventual model as usage-based, with final pricing to be defined after the company sees how customers use the early access service.

The product move comes after years in which language support inside contact centers depended on either interpreters or overlay software.

No Jitter's account notes that LanguageLine introduced an app-based video remote interpreting service in 2013, yet video represented 3% of its revenue by 2015 while telephone interpreting accounted for 86%.

Real-time translation later became more available in text channels during the late 2010s and moved more deeply into agent desktops by the early 2020s.

Voice translation has already appeared through point-solution providers such as Krisp and Sanas, but those deployments commonly require a partnership, a direct commercial relationship or stand-alone software on the agent machine, with audio routed through a virtual microphone layer.

Sanas has a Chrome-based option for browser CCaaS environments, but Market.us Scoop puts Google Chrome thin-client use at roughly one-fifth of contact center agent desktops worldwide.

Embedding the function inside Zendesk changes the implementation path.

Enterprises and BPOs that use SDKs from specialist providers often need custom work from internal IT teams or vendor professional services.

Zendesk's version puts translation into the same environment as routing, contact attributes, call events, controls and recordings.

That context is meant to support smoother handoffs.

A Mandarin-language interaction with Zendesk voice AI, for example, could hand an English summary to a live agent; the stored language attribute could then activate two-way translation automatically.

For administrators, the next documented step is the October early access program.

Its results will shape the language expansion, usage model and commercial terms ahead of the planned Q1 2027 release.