Speech, Music & Hearing > Jens Edlund > Curriculum Vitae
This is a personal webpage - more information

Jens Edlund's Curriculum Vitae

This is a long-ish CV version I use to cut bits out of for various purposes. I've been meaning to sort something less unwieldy in time, but that does not seem to happen.

About me

I was born in Stockholm in 1967. Stockholm has remained my home for the most part: I grew up here, I have spent most of my adult life here, and I currently live here. My first 10 or so professional years I co-managed a small family business in international trade, whilst dabbling with studies in linguistics and other topics as a pass-time. In the mid 90s I changed carreers and went into computational linguistics and speech technology full-time, and have gained about 15 years of research experience in these areas since.
My research interests are broad in scope, with human behaviour and human communication as the common denominator. For the past ten years I've been working with spoken dialogue systems, mostly by first attempting to ascertain experimentally how people behave in human-human interaction, and then to model this behaviour in the systems. I have partaken in the development of several spoken dialogue systems at Telia Research (the former name of Swedish telecom TeliaSonera's research department), at SRI International in Cambridge, and at and KTH in Stockholm (e.g. Telia travel booking service, SLT, AdApt, Higgins, the KTH Connector, and the KTH Reminder).
Having spend my recent past examining ways to make talking computers more responsive and human-like in the way they interact with people, for example by improving their turn-taking skills or by allowing them to use and understand feedback and backchannels, I currently spend my time collecting data that forms the foundation for models of human conversational behaviour on the one hand, and analysing such data on the other hand. I am particularly interested in the interactional aspects of conversations - the way the conversation billows and changes character as it goes along, and the phenomena that govern these changes. I still spend a fair deal of time investigating under what circumstances human-likeness is a good idea in a talking computer, and when it is superfluous or perhaps even a hindrance. Taking the opposite perspective, I'm interested the extent to which basic research can benefit from implementing models and theories in spoken dialogue systems.


I hold a Master of Arts in General Linguistics with specialization in Computational Linguistics from Stockholm University and a PhD in Speech Communication from KTH Royal Instutute of Technology.
Other studies include upper secondary school education in Natural Sciences with focus on Computer Science at Lötskolan in Sundbyberg, Sweden (1983-1986), followed by various courses at Stockholm University including Practical and Theoretical Philosophy, English, and Phonetics (1986-1996).

Work experience


I'm currently working in the speech group at the Department of Speech, Music & Hearing at KTH, and have been since 1999. I currently split my time between data collection in projects such as Spontal (Multimodal database of spontaneous speech in dialog; Swedish: multimodal databas över spontantal i dialog) and investigations of human-human interaction in projects such as Prosody in conversation (Swedish: Samtalets prosodi) and . There are somewhat more detailed listings of my current projects at KTH and my completed projects at KTH avaliable.


Three months of project work at SRI Cambridge. Work consisted of running experiments using the WWW as a corpus for training ASR language models and of Swedish treebanking to a smaller extent.


Project work and consultancy at Telia Research (currently TeliaSonera Sweden AB) related to the SLT (the Spoken Language Translator), and in particular to the Telia travel booking service. Amongst other things, I built the bridge between the Telia travel booking service spoken dialogue system and the Amadeus global distribution system that provided the dialogue system with real live air travel data.

1994 - 2000

Sporadically, I've worked as a consultant within computational linguistics and speech technology (e.g. corpus work for Hapax AB, Stockholm University, and STTS), web design and maintainance (e.g. KTH, Stockhom University, Everyday, Restaurang Trädgårdsvillan), and translation (technichal writing from English to Swedish; e.g. Pearson Education, Peachpit Press).

1984 - 1996

Full-time work in Ola Edlund AB, a family-run company in foreign trade; part-time studies.


I have published a number of articles and book chapters on speech technology in peer reviewed conference proceedings, books, and journals, as well as technichal reports and EU project deliverables. An almost complete list of publications is available.
I have also co-authored a large number of project proposals, both international and national, many of which have received funding. Recent examples include the Swedish Research Council projects Conversational speech synthesis, Multimodal database of spontaneous speech in dialog (Spontal), What makes conversation special?, and the Rhythm of Conversation, the Riksbankens Jubileumsfond project Prosody in Conversation, and the EU projects IURO and GetHomeSafe.
I am a frequent reviewer of journal and conference articles, including articles from the Journal of Speech Communication, Computer Speech and Language, Language and Speech, ACM TiiS and conference and workshop papers from Interspeech, NAACL-HLT, SIGdial, EMNLP, SemDIAL, ICMI (the International Conference on Multimodal Interfaces), MLMI, ECAI and UBICOMP.
I've been the organizer or co-organizer of a number of workshops, conferences and events, including Semdial 2009/DiaHolmia; the Swedish Dialogue Workshop in 2005 (the initial Swedish Dialogue Workshop), 2009 and 2012; the KTH Bullet Course in 2011; the ICT Overlap workshop in 2011; the LREC workshop on Multimodal Corpora in 2012; and the 2012 Interdisciplinary Workshop on Feedback Behaviors in Dialog.
I regularly give invited seminars and talks. Recent examples include a seminars at SSKKII in Gothenburg, Fastnet/Trinity College in Dublin, Voice Provider in Stockholm, CMU Silicon Valley (combined invitation from CMU Silicon Valley and Honda Research Institute), Stockholm University, and a keynote speech at IVA 2011 in Reykjavik.
I spend most of my time in research projects, whch means I do only a very limited amount of teaching. I have co-supervised several exam works and am currently the co-supervisor of one PhD student. I have also taught several summer schools (VISPP Summer School in Kuressare, Estonia, 2008; COST 2102 Training school in 2010; and the CLARA Summer School on Semantic and Multimodal Annotation in 2011). In 2012, I teach three courses at the University of Bielefeld as an invited guest lecturer.
My research interests come together under the heading of face-to-face spoken dialogue, and particularly in human-like spoken dialogue systems. I have done work within ASR (e.g. language modelling; error handling techniques), interpretation and semantics (e.g. domain modelling; robust parsing), dialogue management (e.g. error handling, turn-taking, responsivity), language generation (e.g. generic output formalisms, multi-modal generation), and synthesis (e.g. usage of prosody in speech synthesis, animated talking heads in human-comuter dialogue), and overall spoken dialogue system architecture (e.g. modularity, incrementality, responsivity). I have a particular interest in methodologies allowing studies into the aspects of dialogue that are particular to human conversation.


I have dabbled with computer programming since the beginning of the 80s (a Commodore VIC-20 in 1981 is the first computer I seem to remember handling), and I learned programming in a somewhat more formal manner in upper secondary school on a Compis - a computer put together with the sole purpose of being a training computer for schools, with a programming language (Comal) put together especially for the same reason. Needless to say, this had little to do with any real-life programming I have since encountered. My formal programming training ends with me majoring in computational linguistics. Apart from that, it's all practice.
In my research, my programming skills are a means to an end rather than a subject in itself, and so I'm leaning more towards rapid prototyping and flexibility than towards code optimisation and formal elegance. Hence, I am most proficient in scripting languages, such as Perl and Tcl/Tk, and other medium-to-high level formalisms, such as XML-based technologies, SQL, etc. I also have a working knowledge of a number of other languages and a somewhat more profound understanding of some of the languages and techniques often used in NLP (Natural Language Processing), such as Prolog and machine learning.
I've got hands-on experience with all parts of dialogue system implementation, such as ASR (Automatic Speech Recognition), NLU (Natural Language Understanding), dialogue management, text generation and speech synthesis. Other areas where I have solid experience include text processing on large corpora, language modelling, and communication protocols and data structures, both essential to the kind of distributed applications I generally use in my work. I also have considerable experience with various XML technologies, from HTML and XML to Schemas and XSL transformations, and I have designed and maintained a number of web sites.


I have a long-standing interest in the workings of computer games and internet communities, and combinations of the two. I was somewhat involved in the BBS communities of the late 80s and have been playing on-line community games since sometime before 96, although only very occasionally in the last 8 years. My first real encounter with the internet was as a beta tester for DN Online, an early prototype online version of DN, a large Swedish newspaper. This wasn't really an internet site, but rather a BBS with enhanced graphics provided by something called First Class, but it provided bridges to the real internet and also showed very clearly that proprietary, AOL-style home-spun mini-versions of the internet wasn't the way to go. In any case, the place of language and speech technology in these areas is, to me, a fantastic research area that I have not yet had the opportunity to work with much.
I am also a great fan of information sharing in general and sharing of research materials and results in particular. This has resulted in my participation in various projects related to the accessibility of information:

The Creolist Archives

In 1997, Mikael Parkvall and I started the Creolist Archives (CA; originally the Creole Database Project). CA was a project and a website dedicated to collecting and making public all manner of data relevant to creole and pidgin research. In late 2002, the workload of managing the site finally became overwhelming to Mikael and me and we were forced to take it down. During its 6 year run, the site was used heavily by thousands creolists and linguists from all over the world. Parts of the Creolist Archives, in various incarnations, are saved for posterity by the Wayback Machine.

Web Standards

Since my brief spell as a test pilot for DN's in the mid 90s I'm a keen user of the Web. Like many others, I saw a great need for standardisation if the Web was going to be all that it could be. As a result, I was a vivid proponent of Web standards for a number of years. Amongst other things I took an active part in WASP, "a grassroots coalition fighting for standards which ensure simple, affordable access to web technologies for all". WASP's first priority was to inform developers and consumers of web technology of the merits of web standards, by word-of-mouth as well as by compiling documents such as the WASP Educational FAQ. Luckily, the mission was successful, allowing WASP and others to go on to bigger and better things.

Open Source code

I'm a great friend of open source code. In many cases I am prohibited from releasing sources or even compiled code by license agreements, but whenever I am not and I have time to provide reasonable documentation, I publish my code. I'm currently working on a SourceForge code repository where I'll gather code for speech technology applications and research.
Last updated?May 26th, 2010
Published by Jens Edlund
Copyright © 1998-2010 Jens Edlund