BERKELEY, Calif., May 29, 2024 – Krisp, the world’s leading AI-powered voice productivity software, announced today general availability of its on-device Speech-to-Text (STT) SDKs. Krisp STT technology runs on the end-user device, delivering accuracy and privacy without the need of expensive servers or connectivity. This breakthrough technology has been available within the Krisp application for Contact Centers and Enterprises, and is now available for integration into communications applications via SDKs.
How it works: Krisp STT employs noise-robust deep learning algorithms for real-time on-device speech-to-text conversion. The process consists of several stages, including processing and turning speech into unformatted text, adding punctuation, capitalization and numerical values, redacting PII/PCI, and removing filler words on-device, in real-time. It then assigns text to speakers with timestamps before securely transmitting the encrypted transcript to a private cloud within each application.
“The success of our on-device transcriptions within the Krisp app has fueled demand for our STT SDKs,” said Robert Schoenfield, EVP of Licensing and Partnerships at Krisp. “Krisp STT SDKs are uniquely suited for applications and devices that require secure and accurate on-device transcriptions without the need of servers or an internet connection.”
Since introducing Transcriptions and Meeting Notes within the Krisp application, end users have generated more than 50 million hours of transcriptions. Today, Krisp STT SDKs are available for English and are speaker-accent robust, with plans to add other languages later this year including French, Spanish, and German. Krisp STT SDKs are also noise robust, as the underlying model is trained with noise represented in many use cases.
Krisp STT SDKs are available today for Windows, Mac and Linux, with support for browser-based applications via WASM JS becoming available in the second half of the year. Krisp SDKs deliver industry-leading performance on-device while consuming minimal CPU and memory resources.
About Krisp
Founded in 2017, Krisp pioneered the world’s first AI-powered Voice Productivity software. Krisp’s Voice AI technology enhances digital voice communication through audio cleansing, noise cancelation, accent localization, and call transcription and summarization. Offering full privacy, Krisp works on-device, across all audio hardware configurations and applications that support digital voice communication. Today, Krisp processes over 75 billion minutes of voice conversations every month, eliminating background noise, echoes, and voices in real-time, helping businesses harness the power of voice to unlock higher productivity and deliver better business outcomes.
Learn more about Krisp’s SDK for developers.