Daniel Neiberg, Ph. D
Research engineer

This is a personal web page. More information.

My field of interest is broad and cover spoken dialog, especially paralingustics, but also acoustic to articulatory inversion, inferring speaker identify or cognitive state from voice (e.g. emotion recognition and speaker verification) as well as the neurocognitive machinery underlying these processes. My perspective on spoken dialog can be loosely defined as inferring the dialog state from paralingustic cues, i.e. fundamental frequency, intensity, speaking rate and voice quality. This leads to the following research questions:

• Turn-taking: Who is about to speak next?
• What are the communicative functions of brief feedback?
• What emotional state do interlocutors display?

My thesis is titled "Modelling Paralinguistic Conversational Interaction: Towards social awareness in spoken human-machine dialogue" and cover the state questions above. An erratum is available and a second revision is underway.

Two personal projects are:

• a chess engine in Java. It is quite old now, but it included:
• Bitboards
• Iterative PV search + MVV-LVA + iller moves + history heuristic
• Quiescence search + simplified SSE
• Null-move pruning
• Zobrist hashing
• Positional evaluation function
• Stock market prediction. Since historical data is expensive for high-frequency algorithms, I play around with day-by-day data.

KTH - Royal Institute of Technology
TMH - Dept of Speech, Music and Hearing

Contact information:

Lindstedtsv. 24
100 44 Stockholm, Sweden
Work phone: (+46 8) 790 7567
Email: neiberg[replacewithat]speech.kth.se