Krisp expands CX tools with voice translation SDK

Krisp, a provider of real-time voice AI technology, has launched its Voice Translation SDK, enabling customer experience (CX) platform developers to integrate real-time multilingual voice-to-voice translation into live customer conversations at scale. The technology has been deployed in production CX environments since 2025 as part of Krisp’s Call Center AI platform, operating in live customer interactions globally prior to the SDK’s standalone release.

Unlike text-based translation tools or offline speech processing, real-time voice translation operates on continuous audio streams, where latency, accuracy and conversational flow are closely interdependent. Systems must recognise a wide range of accents, function reliably in noisy environments and preserve natural turn-taking. Reducing latency excessively can affect context and increase errors, while longer processing times may disrupt the flow of conversation.

The Voice Translation SDK is designed to balance these constraints in live, two-way interactions. It supports combinations of more than 60 languages and is optimised for synchronous conversations where clarity and continuity are essential. This allows multilingual exchanges within live conversations without the need for human interpreters, which is particularly relevant in customer experience settings requiring immediate language support.

Davit Baghdasaryan, Co-Founder and Chief Executive Officer of Krisp, said that language barriers in global customer experience operations can affect response times and customer satisfaction. He noted that real-time voice translation must function reliably in large-scale production environments and that making the technology available as an SDK enables CX platforms to embed multilingual voice capabilities directly into live systems.

The SDK is available for Windows, macOS and web-based applications, supporting integration into both native and browser environments. To improve performance in real-world conditions, Krisp applies local noise cancellation before audio is processed in the cloud, helping to isolate the primary speaker and enhance recognition accuracy. The SDK also supports custom vocabulary and domain-specific dictionaries, enabling organisations to maintain consistent terminology in specialised contexts.

The Voice Translation SDK extends Krisp’s broader portfolio of voice AI models for customer experience, which includes noise cancellation, accent conversion and voice isolation for Voice AI Agents (VIVA), designed to reduce false interruptions and improve turn-taking. According to the company, its technology is used across more than 200 million devices and processes over 80 billion minutes of voice conversations each month.