News

Publication on far-field automatic speech recognition in social media

The Februar 2021 issue of the prestigious journal  "Proceedings of the IEEE" published an article entitled  "Far-field automatic speech recognition" by R. Haeb-Umbach and coauthors. This publication is now promoted by the IEEE, the world's largest technical professional organization for the advancement of technology, in social media  (LinkedIn, Facebook, Twitter). The links can be found here:

    LinkedIn
    Facebook
    Twitter

In this article, R. Häb-Umbach and the coauthors L. Drude and J. Heymann (Amazon), M. Delcroix and T. Nakatani (Nippon Telephone & Telegraph, Japan), as well as S. Watanabe (Carnegie Mellon University, USA) describe the speech technology for far-field speech recognition, that is for situations where the distance between the speaker and the recording device is large. This scenario, which is the typical usage scenario of digital home assistants, such as Amazon's Alexa, is characterized by a severe degradation of the speech quality caused by room reverberation, noise and other acoustic distortions (e.g., tv or music in the background). This requires special measures to ensure reliable speech recognition, which are described in the aforementioned journal article.

Contact