Mavenir Voice AI is moving artificial intelligence from standalone apps into the telecom network itself through a new partnership with real-time Speech AI company Sanas.
The companies plan to integrate Sanas technology into Mavenir’s MAVcore Voice AI portfolio, enabling operators to deliver real-time speech enhancement, accent transformation and language understanding directly through mobile core networks. The platform is also expected to add deepfake detection and fraud signals designed to identify synthetic-voice attacks.
Why It Matters
Voice AI is typically delivered through an app or external cloud platform. Mavenir and Sanas are taking a different approach by embedding AI capabilities at the carrier level.
This could give telecom operators greater control over latency, privacy and data residency while allowing them to offer translation, enhanced voice quality and security as network services. Mavenir is also positioning Voice AI as a potential source of new operator revenue rather than simply another network cost.
Korean Market Context
The model could be particularly relevant to South Korea, where advanced 5G networks and strong adoption of AI services create favorable conditions for network-native AI.
Korean carriers are already expanding beyond traditional connectivity into AI platforms and enterprise services. Carrier-level translation or synthetic-voice detection could provide another layer of differentiation, particularly for international calls, customer service and fraud prevention.
Market and Competitive Context
Mavenir competes in the IMS and voice-core market with major telecom infrastructure suppliers including Ericsson, Nokia and ZTE. A 2026 GlobalData assessment cited by Mavenir placed MAVcore among the industry’s top-rated platforms.
The company has also been developing network-based AI services beyond this partnership. Its 2026 demonstrations include real-time two-way voice translation and AI agents running directly through telecom infrastructure.
Sanas brings specialized real-time speech technology to that strategy, potentially allowing operators to add intelligence to conventional phone calls without requiring both users to install the same application.
Company Claims vs. Verified Facts
The partnership itself and the planned integration of Sanas into MAVcore are confirmed by official announcements from both companies.
However, there is an important distinction: the joint service is not yet commercially available to operators. Mavenir and Sanas say they will begin joint technical validation before commercial availability follows.
Claims that the technology will lower operating costs, create new revenue or provide effective protection against deepfake fraud therefore remain expected benefits rather than independently demonstrated commercial results.
What to Watch Next
The biggest challenge will be performance at telecom scale. Translation needs extremely low latency, speech transformation must remain natural, and deepfake detection must minimize false positives during real calls.
Carrier adoption will provide the real test.
If operators can successfully deploy these capabilities while keeping sensitive voice data within their networks, Mavenir Voice AI could represent an important shift: AI would become part of the communications infrastructure itself rather than another application running on top of it.
