Johan Andreas Rohdin

JOHAN ANDREAS ROHDIN

PROJECT: SEQUENce Summarizing neural networks for speaker recognition

Host Institution: Faculty of Information Technology, BUT BRNO

Country of Origin: Sweden
Previous Workplace: Japan

TOPIC OF MY RESEARCH:

Topic of my research deals with use of neural networks for speaker recognition.

WHY I HAVE CHOSEN THE TOPIC:

I have long-term experience in this research topic and publications and FIT BUT is also a
world-wide known centre in this sphere. I am looking forward to share the knowledge together and I believe we achieve outstanding results.

USE OF THE PROJECT RESULTS IN PRACTICE:

Extracting information from the speech is interesting from the researcher´s point of view - there is a connection between mathematics, mechanical engineering, "big data" and language formation learning - and I am happy that the research will be used for practical applications. The main business customer of our results will be the company Phonexia which has been founded by the BUT "speech recognizers" in 2006 and which has grown into the significant world player in speech technologies. However, Phonexia does not have a monopoly on our work, we publish our results and they are publicly available. We will cooperate also with other European and world companies.

WHY SOMOPRO:

I have finished my Ph.D. in a top research group in Japan and I worked in the local IT company. However, I wanted to return back to Europe as I come from Sweden, to some top-class group working in the field of my dissertation. The group from BUT Brno is very famous and my Japanese supervisor has recommended it. When I learned from them that there is a program SoMoPro to support arrival of foreign researchers, the choice was clear. 

You can read more about the project HERE .