Within the newly funded project SCRIBE, the University of Science and Technology vacancies for two PhD Research Fellow positions at the Department of Electronic Systems (IES). The PhD positions are for up to 4 years, with 25% work assignments for NTNU IES / IDI.
|Scholarship Sponsor||Norwegian University of Science and Technology|
|Scholarships level||PhD Position|
|Award Amount||NOK 482 200 per annum|
|Fellowship Period||Four years|
|Study area||Electronic Systems|
|Opening date||February 25, 2021|
|Closing date||March 23, 2021|
At NTNU, creating knowledge for a better world is the vision that unites our 7,400 employees and 42,000 students. We are looking for dedicated employees to join us. You will find more information about working at NTNU and the application process here.
About the position
You will report to the head of the position.
Duties of the position
The successful candidates will work with different aspects of modelling conversations:
Modelling conversational speech
Conversational, multi-party speech contains false starts; repeats; incomplete and/or ungrammatical sentences, non-verbal sounds; overlapping speech; interruptions, etc. Such phenomena significantly impact both acoustic and language models in an address recogniser. The lack of (rich) transcriptions of conversational speech cause language models also for conversational ASR to be trained on large corpora of written text, with a resulting mismatch to the task domain.
The introduction of attention mechanisms, in particular “transformers”, has pushed the state-of-the-art in many natural language processing (NLP) areas. Lately, ASR models have also incorporated transformer techniques both in acoustic and language models with excellent results.
This PhD project will address the modelling of conversational speech from small amounts of relevant training data. Issues like transfer learning, transformers, context modelling and semantically meaningful performance measures will be central.
Characterisation and handling of dialects
Conversational speech exhibits pronunciation variations due to its spontaneous nature, different degrees of formality, varying emotional state and speaking rate, as well as variants in dialects and accents. These variations pose challenges to speech recognition in most languages. Transcriptions reacting to acting dialect use are sparse in most languages, including Norwegian. Spoken dialect corpora with corresponding transcriptions are even less available. This is a severe hindrance to the adequate modelling of dialects.
This PhD project will use recent machine learning techniques to separate linguistic content from dialect variation jointly on the spoken and written utterances. This model will recognise colloquial speech and be interpretable, providing better insights into relevant acoustic-phonetic and linguistic characteristics of Norwegian dialects.
From the latent representations learned by the model, it will also be possible to perform dialect identification given the spoken utterance. This novel model’s performance will be compared to two more standard paradigms, namely end-to-end modelling (powerful but non-interpretable) and explicit dialect modelling (inefficient at using scarce data resources).
This work will be carried out in collaboration with Telenor Research researchers and from the Computer Science department at NTNU. The candidates will also contribute to the definition of meaningful metrics to evaluate conversational speech recognition systems and improve linguistic resources at the Norwegian National Library (NB) and the Norwegian Broadcasting Corporation (NRK) that our project partners. They may also collaborate with parallel research projects in the domain of machine learning and speech technology already running at NTNU.
The deadline for applications is March 23, 2021
We seek a highly-motivated individual who has
- Strong background in machine learning, signal processing and mathematics with a research-oriented master thesis in a related field, e.g., statistical machine learning, artificial intelligence, statistical signal processing, speech technology, information theory, applied mathematics, or optimisation
- Experience with programming
- Good written and oral English language skills
Publication activities in the disciplines mentioned above will be considered an advantage but is not a requirement. Knowledge of a Scandinavian language is regarded as a plus.
The PhD position’s main objective is to qualify for work in research positions. The qualification requirement is that you have completed a master’s degree or second degree (equivalent to 120 credits) with a strong academic background in machine learning and signal processing or equal education with a B or better grade in terms of NTNU’s grading scale.
If you do not have letter grades from previous studies, you must have an equally good academic foundation. If you cannot meet these criteria, you may be considered only if you can document that you are particularly suitable for education leading to a PhD degree.
- Challenges the status quo and promotes new initiatives
- Is good at anticipating problems and identifying logical solutions as well as contradictions and inconsistencies
- Acquires new knowledge quickly and can use existing knowledge in new ways
- Works together with others across established reporting lines
- Exciting and stimulating tasks in a healthy international academic environment
- An open and inclusive work environment with dedicated colleagues
- Favourable terms in the Norwegian Public Service Pension Fund
- Employee benefits
Salary and conditions
PhD candidates are remunerated in code 1017 and are usually remunerated at gross from NOK 482 200 per annum before tax, depending on qualifications and seniority. From the salary, 2% is deducted as a contribution to the Norwegian Public Service Pension Fund.
The period of employment is four years, with 25% work assignments for NTNU IES.
Appointment to a PhD position requires that you are admitted to the PhD program in [subject area] (Link to website, if applicable) within three months of employment and participate in an organised PhD program during the employment period.
The engagement is to be made following the regulations in force concerning State Employees and Civil Servants and the acts relating to Control of the Export of Strategic Goods, Services and Technology. Candidates who assess the application and attachment are seen to conflict with the latter law criteria will be prohibited from recruitment to NTNU. After the appointment, you must assume that there may be changes in the area of work.
The position is subject to external funding.
It is a prerequisite you can be present at and accessible to the institution daily
The application and supporting documentation to be used as the basis for the assessment must be in English.
Publications and other scientific work must follow the application. Please note that applications are only evaluated based on the information available on the application deadline. You should ensure that your application clearly shows how your skills and experience meet the criteria set out above.
The application must include:
- CV, certificates and diplomas
- Academic works – published or unpublished – that you would like to be considered in the assessment.
- Name and address of three referees
Please submit your application electronically via jobbnorge.no with your CV, diplomas and certificates. Applications submitted elsewhere will not be considered. A diploma Supplement is required to attach for European Master Diplomas outside Norway. Chinese applicants are required to confirm the Master Diploma from China Credentials Verification (CHSI).