Vodafone and Ericsson have completed a trial of two AI-powered voice services — real-time noise cancellation and live language translation — delivered directly from the mobile network rather than through an app on the handset.

The trial sits inside the operators' joint "Core Network of the Future" programme, which examines how AI can be embedded directly into telecom networks. Because the services run in the core, customers need neither a separate application nor a new device to use them. The technical setup pairs Vodafone's application server and AI engines with Ericsson's IP Multimedia Subsystem (IMS), the signalling layer that handles voice and multimedia sessions over LTE and 5G networks.

The first service uses AI to identify and strip out background noise during a live call. Vodafone positions it as useful for customers working or travelling in noisy environments, improving call quality without specialist hardware. The second service translates between two different languages in real time during a phone conversation — a proposition aimed squarely at businesses that support customers, suppliers and colleagues speaking different languages.

The commercial logic behind both is straightforward. Voice has long been commoditised, and operators have watched over-the-top services absorb innovation at the application layer. Moving AI-driven features into the core reverses that pattern: the operator network itself becomes the differentiator, and any device on it can access the capability.

Marco Zangani, Vodafone's Director of Network Strategy and Architecture, framed the trial in historical terms. "AI is giving the mobile network back its voice. More than 50 years after the first mobile call, innovation is accelerating and shifting back into the network," he said. "Over the past few years, we have built the infrastructure, data capabilities and safeguards needed to unlock AI for our customers to converse more clearly."

Nadine Allen, Head of UK and responsible for Vodafone Group at Ericsson, described the trial as evidence of what network-native AI can do. "By integrating Vodafone's powerful AI capabilities with Ericsson's core network and media processing technologies, we are proving that sophisticated services like real-time translation and noise cancellation can be delivered seamlessly," she said. "This is a crucial step toward creating more intelligent, programmable networks that unlock new value for customers."

Both statements are vendor-operator marketing claims until the services reach commercial availability, and neither company has disclosed deployment dates, pricing or target markets in the trial announcement. What the trial does establish is technical feasibility: the AI processing can run within the IMS architecture Vodafone already operates, rather than depending on endpoint hardware or third-party apps.

The work also fits the broader standards direction in voice services. IMS already underpins VoLTE and VoNR, and operators are progressively upgrading their voice cores ahead of 5G standalone rollouts — a context in which programmable, AI-capable network functions become easier to justify commercially.

The partnership has recent form. VodafoneThree, the merged UK operator, has activated a new Ericsson 5G core, and Nokia and VodafoneThree have switched on their first 100 advanced 5G radio sites, signalling that the group's core modernisation is well under way.

For now, noise cancellation and live translation remain trial-stage capabilities. Whether they graduate into revenue-generating services will depend on how quickly Vodafone can move them from its "Core Network of the Future" programme into live networks across its European footprint.