Radisys Launches In-Call Speech Analytics Integrating Sensory's TrulyHandsfree
Radisys, a provider of open telecom solutions, has integrated Sensory's TrulyHandsfree Voice Control speech recognition software into its Engage Media Server. This integration enables service providers to deploy a wider range of e-commerce, customer service, and unified communications (UC) applications with speech interfaces.
Radisys' Engage Media Server is a cloud-native media server platform for voice and video communications, content processing, and media quality optimization.
The integrated solution for voice commands and wake word detection enables the creation of new applications for in-call speech processing. It works seamlessly with Radisys' support for conversational speech-to-text, text-to-speech, and speaker verification, providing a suite of speech processing tools that allow developers to build new services, including a combination of wake word, command set, conversational speech-to-text, and speech transcription features, within any application.
Radisys' solution is integrated with media servers already deployed in service providers' networks, eliminating the need forseparate, dedicated network elements for speech processing. It can be deployed in public or private clouds or on premises.
The integration of Sensory's TrulyHandsfree Voice Control offers multiple-phrase technology that recognizes, analyzes, and responds to dozens of keywords, even consistently recognizing phrases when they are embedded in sentences or surrounded by noise. This new offering claims more than 95 percent accuracy.
"Historically, in-call speech processing has taken a one-size-fits-all approach [with] conversational interfaces that convert natural language into text," said Al Balasco, vice president of communications solutions at Radisys, in a statement. "These are expensive in terms of hardware and software costs, and overkill for many applications where there is a small set of commands like record call, mute line, or start video, or where more sophisticated conversational speech-to-text can be initiated by a wake word before the conversational engine begins processing. Not only does this approach save money, it makes a whole range of valuable new services viable."
"In-call speech recognition has traditionally been expensive, resource-intensive, and highly complex overall to implement. Additionally, there are numerous challenges to ensuring peak speech recognition performance and accuracy, notably varying network conditions and call quality. Our technology combined with Radisys' powerful Engage Media Server changes the playing field because it offers simple, extremely accurate phrase-spotted command and control functionality that delivers an efficient voice assistant UX, all on a single platform," said Bernie Brafman, vice president of business development at Sensory, in a statement. "We're very pleased to be teaming with Radisys on this valuable solution for service providers and application developers."