Research Associate

The University of Edinburgh - College of Science and Engineering - Informatics

The School of Informatics, University of Edinburgh, invites applications for a full-time, fixed term Postdoctoral Research Associate position to implement different models of unsupervised and weakly supervised representation learning from speech, and to evaluate and compare these models, especially with respect to human perceptual data from the literature.

The project is jointly funded by US and UK agencies, with co-PIs Naomi Feldman (U Maryland) and Sharon Goldwater (U Edinburgh). We plan to build on ideas from the zero-resource speech processing literature and focus on the ways in which these models do or do not capture human perception (e.g. the development of native language speech perception and/or perception of non-native speech sounds by adults). We are particularly interested in the extent to which top-down information is needed for representation learning, and if so what type of top-down information (e.g. knowledge of phonetic categories or words). We plan to work with corpus data from a variety of languages.

One postdoc will be hired at each site, and close coordination is expected, with occasional visits by the researchers to each other’s labs. The Edinburgh team will be primarily responsible for creating the learning models, and the Maryland team will be primarily responsible for creating the evaluation models that will link the output of the models to data from human perceptual experiments.

Please ensure your application includes an up to date CV.

Informal enquires should be direct to Sharon Goldwater (

This is a full time, fixed term position from 16 April 2018 (or as soon as possible thereafter) for a period of 36 months.

This is a re-advertisement and previous applicants need not apply.

The closing date is 5pm on Wednesday 17 January 2018.

For further particulars and to apply for this post please click on the 'apply' button below

Share this job
  Share by Email   Print this job   More sharing options
We value your feedback on the quality of our adverts. If you have a comment to make about the overall quality of this advert, or its categorisation then please send us your feedback
Advert information

Type / Role: