Improved Speech Intelligibility and Automatic Speech-to-Text Conversion for Call Centers

Press Release /

At Call Center World (CCW), taking place February 21 – 23 in Berlin, Fraunhofer IDMT will be presenting state-of-the-art software solutions for improved speech intelligibility and automatic speech-to-text conversion.

© Thinkstock/Wavebreakmedia Ltd
Thanks to the Fraunhofer solution for improved speech intelligibility users can adjust sound and volume to their individual hearing needs.

The work in call centers is characterized by high levels of ambient noise. Under such difficult acoustic conditions, focusing on the telephone conversation with the customer and simultaneously retrieving information from various databases requires utmost concentration from the call center agent. At CCW, the largest international conference and trade show for call center management in Europe, the Fraunhofer experts from the Hearing, Speech, and Audio Technology project group will be presenting two solutions offering effective support in the everyday work of call center agents.

Fraunhofer’s system for improved speech intelligibility can easily be integrated with any telephone system and headset. Over a user interface of the PC or end device, the user may select from a number of preset signal outputs the one deemed most convenient in terms of sound and volume. The signal of each incoming call can then be adjusted to the user’s individual hearing needs. In addition, background noise interfering with the telephone conversation is filtered out from the signal. »At the heart of the system is a very intelligent signal processing algorithm, which allows significant improvement of speech intelligibility, regardless of whether the signal is of good or poor quality«, says Dr. Jan Rennies-Hochmuth of Fraunhofer IDMT. »The system enables also users with an impaired hearing ability to conduct effective telephone conversations, even in noisy environments«.

The second solution to be presented by Fraunhofer allows automatic conversion of oral conversations into written text, including search for keywords. »The idea behind this solution is that during the conversation the call center agent is being offered data and information assisting in the customer care process. But the system is also able to automatically identify the agent most capable of processing the request, who may then take over the call«, explains Jan Rennies-Hochmuth. As a highly beneficial side effect, the system helps meet the obligation for call center operators to document all telephone conversations between their agents and the customers. Compliance with data privacy regulations, which is another obligation to be kept by call center operators, is ensured at any time, as the speech recognition software runs on the computers of the call center (i.e., no data is stored in the cloud). 

Please feel free to stop by Booth C11 in Hall 2, and get to know Fraunhofer’s state-of-the-art solutions for improved speech intelligibility and automatic speech-to-text conversion. 

About the Project Group Hearing, Speech and Audio Technology of the Fraunhofer Institute for Digital Media Technology IDMT

The goal of the Project Group for Hearing, Speech and Audio Technology of the Fraunhofer IDMT is to implement scientific findings on auditory perception of normal and impaired hearing in technological applications. Scientists carry out applied research and development on behalf of industrial companies and public institutions in the fields of telecommunications, multimedia, health and care services, building technology, transport, industrial production and security. The project group was established in Oldenburg in 2008 as a branch of the Fraunhofer Institute for Digital Media Technology IDMT. Through scientific cooperations, it has close links with the hearing research facilities in Oldenburg and is also partner in the cluster of excellence »Hearing4all«.