Vodafone tests live network AI translation with Ericsson

Vodafone and Ericsson have completed a live network trial running AI translation during standard voice calls across five European languages.

The test evaluated direct audio interpretation between English, French, German, Italian, and Spanish speakers. Engineering teams also tested active noise suppression across the carrier channels. The processing isolated human speech and removed heavy acoustic interference typical of crowded railway stations.

Marco Zangani, Director of Network Strategy and Architecture at Vodafone, said: “AI is giving the mobile network back its voice.”

Zangani noted that more than five decades have passed since the first mobile telephone call. While computing tasks drifted toward personal devices in recent decades, processing power is returning to carrier infrastructure. 

Over several years, Vodafone built the physical data facilities, computational pipelines, and administrative safeguards necessary to host speech models directly within transit switching layers.

Core network integration bypasses device software

Network-level execution alters the delivery mechanism for conversational services. Traditional language translation and audio-cleaning tools demand standalone consumer applications or specialised handset silicon. 

Under this carrier architecture, voice streams route through telecom processing nodes. Subscribers place ordinary telephone calls on existing and unmodified handsets. Because audio manipulation occurs upstream within the core infrastructure, terminals receive clean translated voice streams directly.

The deployment links proprietary components from both corporate partners. Vodafone contributed its internal application server and an in-house AI platform. Ericsson integrated an AI gateway with its core network infrastructure and media processing environments.

APIs connect the telecoms transport layer with the AI domain. These interfaces route real-time data exchanges between call switches and machine learning models.

This infrastructure division yields tangible advantages. Callers experience no device battery penalties or local processing lag. In addition to the current European languages, network engineers confirmed that the underlying architecture accommodates additional language libraries as commercial needs dictate.

Vodafone and Ericsson’s joint initiative coordinates international service deployment

The voice trials form a central milestone within the ‘Core Network of the Future’ initiative, a collaborative research framework operated by Vodafone and Ericsson. The programme develops and evaluates next-gen telecoms technologies. Both partners intend to establish the core network as an intelligent operational base that drives new commercial services alongside internal operational efficiency.

Hosting capabilities within central switching centres shortens deployment cycles across diverse territories. Software additions activated at the network level take effect across all active mobile subscriptions simultaneously. This design enables Vodafone to introduce voice features across multiple national operating companies.

Vodafone clarified that the network translation mechanism targets everyday communication needs rather than professional translation sectors. Target applications span cross-border commercial negotiations, overseas holiday logistics, hotel reservations, and dining bookings.

The capability also assists travellers during urgent overseas emergencies. In those moments, rapid language translation allows callers to communicate clearly with local emergency responders.

Nadine Allen, Head of UK and responsible for Vodafone Group at Ericsson, said: “This successful trial with Vodafone is a demonstration of how network-native AI can redefine the mobile experience.

“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.”

Successful completion of these live carrier trials establishes network-native machine learning as a viable architecture for future voice telecoms.

See also: China Telecom AI launches Xing4.0-29B-A4B open model

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the IoT Tech Expo and Cyber Security & Cloud Expo. Click here for more information.

Telecoms is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.