Krisp Launches Real-Time Voice Translation API for Developers

Voice AI company Krisp has launched Voice Translation v3, a new version of its real-time translation platform, along with a self-service Voice Translation API that allows developers to integrate the technology into their own applications.

The company said the platform is designed to improve multilingual communication in industries such as healthcare, financial services, customer support and business process outsourcing (BPO).

According to Krisp, the translation engine supports 61 languages and can translate conversations in real time while handling challenges such as background noise, accented speech and industry-specific terminology.

The latest version introduces several features aimed at enterprise customers. These include automated quality assessment of translated calls, live bilingual transcripts, language auto-selection, custom vocabulary controls and call auditing tools that allow administrators to review conversations from both the agent and customer perspectives.

Davit Baghdasaryan, CEO and Co-Founder of Krisp, said the company developed the technology for environments where translation accuracy is critical.

“Real-time voice translation is growing rapidly, but many systems have not been tested in situations where accuracy is essential. We built our engine for sectors such as healthcare, insurance and financial services, where mistakes can have serious consequences,” he added.

Alongside the enterprise offering, Krisp has also opened access to developers through its new Voice Translation API. The API allows developers to send speech input and receive translated speech and text output through a single connection. JavaScript and Python software development kits (SDKs) are available at launch with additional support planned.

Robert Schoenfield, Executive Vice President of Licensing and Partnerships at Krisp, said that developers building applications in areas such as telehealth, customer support and fintech require translation tools that perform reliably in real-world conditions.

The company noted that the technology has already been used in live deployments. In one healthcare implementation, the company reported that the system handled 90% of multilingual calls without requiring a human interpreter and achieved 96% translation accuracy across more than eight languages.