|Position:||Assistant Professor in Intelligent Systems with spec. in Machine Learning
Division of Speech, Music and Hearing (TMH)
School of Electrical Engineering and Computer Science
KTH Royal Institute of Technology
|E-mail:||ghe at kth se
|Address:||KTH Royal Institute of Technology
Lindstedtsvägen 24, office 518
SE-100 44 Stockholm
My research interests are probabilistic modelling and machine learning for data generation tasks, most prominently statistical speech synthesis but also motion and gesture. My interests extend to the perceptual effects of modelling assumptions and design decisions, statistical robustness, and how to perform and analyse subjective tests.
I have a Ph.D. in electrical engineering (telecommunications) from KTH. Following graduation, I was a post-doc at the Centre for Speech Technology Research (CSTR) at the University of Edinburgh, UK, (2013–2016) then in Prof. Junichi Yamagishi's lab at the National Institute of Informatics, Tokyo, Japan, (2016–2018), and then at Speech, Music and Hearing (TMH) at KTH in 2018, prior to being promoted to my current position in January 2020.
StudentsI am fortunate to be supervising the following doctoral students:
- Taras Kucherenko (main supervisor Hedvig Kjellström)
- Ghazaleh Esfandiari Baiat (main supervisor Jens Edlund)
- Patrik Jonell (main supervisor Jonas Beskow)
A list of my publications is provided on a separate page, grouped by type:
- Books and book chapters
- Journal papers
- Conference papers
- Non-peer reviewed conference contributions
Selected presentation materials
Practical classifier design
A lecture from the M.Sc. course EQ2341 Pattern Recognition and Machine Learning at KTH Royal Institute of Technology, Stockholm, Sweden, most recently given in May 2019.
[ slides ]
Cyborgs and other controllable synthesisers: an update on past and planned research
A presentation given at the Centre for Speech Technology Research, the University of Edinburgh, UK, in November 2018.
[ slides w/o audio | abstract ]
Wagging speech by the tail: The case for robust data generation
A presentation most recently given at the ACCESS Linnaeus Centre, KTH Royal Institute of Technology, Stockholm, Sweden, in October 2018.
[ slides w/ audio (2.6 MiB) | slides w/o audio | abstract ]
Introduction to modern and controllable speech synthesis
Excerpt from a presentation given at the annual EACare project retreat in Sigtuna, Sweden, in April 2018.
[ slides w/ audio (6.6 MiB) | slides w/o audio ]
Perceptual debugging of speech synthesis: Using speech from the future to pinpoint the design decisions that matter most
A presentation most recently given at the Tokuda and Nankaku Laboratory, Nagoya Institute of Technology, Japan, in January 2018.
[ slides w/ audio (7.6 MiB) | slides w/o audio | abstract ]
Note: Some audio files have been removed from the slides and presentations on this page due to strict Japanese regulations regarding personal data. Audio playback from pdf is known to work in Adobe Acrobat Reader (from trusted documents), but not in Preview on macOS.
Gustav Eje Henter, Thomas Merritt, Matt Shannon, Catherine Mayo, and Simon King (2014)
Repeated Harvard Sentence Prompts (REHASP) corpus version 0.5
Studio recordings of 30 Harvard sentences, each read aloud 40 times by a female native speaker of British English. Released under a CC BY 4.0 license.
[ official repository | readme file | .bib | publication ]
Links and additional information
- Download my curriculum vitæ.
- Download a .bib file with all my publications.
- Access my profile on Google Scholar.
- Find my ORCID iD (0000-0002-1643-1054).
- Access my page on GitHub. (Not yet up to date.)
- Access my official profile on KTH Social.
- Disclaimer for personal web pages at KTH.
This page was last updated 2019-12-16.