Idiap has a new opening for 3 Postdoc positions in the areas of multilingual speech processing and cross-lingual indexing

The Idiap Research Institute invites applications for three post-doctoral positions in the general areas of multilingual speech recognition, low-resourced speech recognition, domain adaptation and cross-lingual indexing. The positions are currently funded by EU H2020 and IARPA projects, ranging from one to five years in duration.

The successful candidates will work with Prof. Hervé Bourlard, Dr. Philip N. Garner, and/or Dr. James Henderson, and will have the opportunity to collaborate with some of the 25 members of the Audio and Speech Processing and Natural Language Processing groups, engaged in world-class research in speech and natural language processing. Exceptionally qualified candidates can also be considered for a longer-term Research Associate position.

Description of work

Modern speech recognition requirements are multilingual in nature, representing regions of the world that are of interest to news organisations and governments. Such locations often yield languages that have limited linguistic resources for academic purposes, or use unusual accents and dialects. Recordings are not studio quality; rather, they are obtained in noisy environments over limited channels. To best make use of the the data, it is not only necessary to perform speech recognition, but also provide the results in a manner suitable for machine translation, summarisation and indexing.

Idiap researchers have developed techniques to handle the multilingual nature of communication in Switzerland and abroad, exploiting advanced speech processing, statistical modeling and machine learning. Much of this is based on deep neural networks, but also encompasses sparsity and low rank modelling. This extends into natural language processing using domain specific word embedding and hashing of continuous spaces.

The available positions span the above technology range, from low level signal processing to higher level natural language processing and translation. In the context of international collaborations, the successful candidates will work using these techniques and others with partners in countries as diverse as the UK, the USA, Germany, Latvia and Qatar.


Building upon a strong PhD record, candidates should have a strong background in engineering, mathematics or a related discipline, should be computer-literate, familiar with modern distributed programming environments and languages such as C++, Python and Perl. Familiarity with speech processing tools such as Kaldi and deep learning toolkits such as Torch will be a distinct advantage. An exceptional academic record and a clear aptitude for creative (and independent) research will also be priority selection criteria.


All positions are offered as one-year basis with the possibility of renewal based on funding and performance. The positions will remain open until filled, but all applications received by September 30th, 2017, will be considered.

To apply for this position, click on the following link: 3 Postdoc positions in the areas of multilingual speech processing and cross-lingual indexing