Dear ISCA board,

in appreciation of the new program that supports PhD students to visit undergraduate Universities or other places that do not have a lab or existing research program in the area of ISCA interest, we would like to submit the student reports on their experience.

Mengjie Qian, from Martin Russel’s lab at Birmingham University was able to visit us for one week, approximately a month before the deadline for Interspeech. The topic of interest was the SLaTE hosted competition:

Special session 2: Spoken CALL challenge
This second special session will be focused on the Spoken CALL challenge, jointly initiated by Université de Genève, University of Birmingham and Radboud University.

The task consists of labelling L2 English utterances by Swiss German teens as accepted or rejected in relation to a grammatically and linguistically correct gold standard.

The challenge has been launched to allow quantitative comparisons between methodologies used by different researchers on the same task. Instructions, data and further information

Two of our students successfully participated with the help of University of Birmingham in an exchange similar to this year. This event inspired the proposal for ISCA to support this sort of interaction. One paper was submitted and accepted to SLaTE 2017. This year, the three students submitted three(!) papers to the SLaTE special session and we are happy to report that one of these was accepted and we will be there to present it. The experience has been what they called “hard fun”! The students learned a lot. I believe the submitted systems are interesting contributions and will help these students to think of speech as a viable topic for their Master studies. A third contribution is a tutorial and an open source software kit that may help new students to have a faster start on the topic. Should at least one paper get accepted, this would of course magnify the experience even more.

The following are their statements:

Dominik Jülg, Bachelor student, 6^th semester, Germany:

“The exchange program really brought our project forward as Mengjie helped us to examine which ways we can go. As a beginner, it is really hard to orient oneself in such a versatile topic like speech and language processing. Furthermore, a professional with deeper knowledge can help to answer questions and speed up the development process in general.”

Mario Kunstek, Bachelor student, 6^th semester, Germany:

“Studying at the CSU-K (Cooperative State University in Karlsruhe) includes a student research project so we decide to be part of the Shared Task 2 (hosted by, SLaTE, as mentioned above). While trying to set up the baseline system and preparing the data for it there was a point where support from an expert for the ASR systems was needed.

Mengjie Qian traveled to Germany to teach us about the baseline and support us. Apart from setting up the baseline system, she also worked with us on the CSU-K-DNN and the CSU-K Toolkit. Another topic she taught us was NLP.

Working with her, felt as if we had been a team for a long time. She suited our team very well. The result of this cooperation was that two solutions for the Spoken CALL were developed written up and submitted to Interspeech SLaTE Special Session.

In summary, the cooperation was very helpful and without Mengjie’s support, we never would have reached these results that we now have.

Many thanks to ISCA for making this exchange possible and again, many thanks to Mengjie.”

Cem Freimoser, Bachelor student, 6^th semester, Germany:

“For our student research project, that is part of our scientific training at the CSU-K, the support of an expert for ASR systems was necessary. For a long time during Semester 5, we were unable to make progress with the task because there were technical difficulties in using the Baseline system, which required the experience of an expert for the solution. This support was provided by Mengjie thanks to ISCA. Mengjie not only supported us in setting up the baseline system, but also in developing the CSU-K DNN Classifier and CSU-K Toolkit. She also taught us a lot about NLP. Besides the technical aspects, Mengjie also fit perfectly into our team, resulting in a great group dynamic. So we had very productive discussions in the team about how the CSU-K system can be improved. The cooperation has resulted in two systems that can be used for the Spokencall. This cooperation also made it possible to develop approaches that could be used in future experiments for the CSU-K System. Without the help of Mengjie, I think such good results would have been unthinkable.“

Mengjie Qian, PhD student, University of Birmingham

“The experience of working at Cooperative State University Karlsruhe was fascinating and I learned a lot from the Bachelor students and from Ms. Berkling. Our different background benefited us and brought us many wonderful ideas. As it was a short visit, each of us was concentrating on the task and we got more chance to discuss the problems and ideas face to face. Such convenient communication made it easier to solve problems and helped us implemented the ideas quickly and efficiently. We finally produced satisfactory results for the task and we would not have achieve those so fast without this trip. I also got a chance to see the views in Germany. The students kindly showed me around the city and introduced me a lot about German culture. Moreover, I am glad that I made a few friends during this visit. This is really an enjoyable and fruitful visit.”