Job Description

A Corp Lab in NTU is looking for a Research Scientist or Research Engineer II who will work on developing algorithms using large language models for Automatic Speech Recognition (ASR) tasks with the key goal to develop ideas into publications and patents. The new headcount will also assist in processing speech data, implementing and evaluating deep learning models, and integrating the model with other project components and demo. Relevant experience and knowledge in speech and language processing and well-known speech processing frameworks is a must.Key Responsibilities:Research and develop relevant speech and language technologies to facilitate the project requirements - large language model and speech recognition for AI applications.Conducting regular model training and evaluation for ASR tasks.Regularly reporting experimental results to supervisor and project funders, propose solutions for further improvement and ensure targeted deliverables are met at each project deliverable milestones.Organize and prepare acoustic and text data for text to speech and speech recognition training using end-to-end/deep learning frameworks such as Huggingface, Whisper, ESPnet.Write document and transfer developed models to engineering team and/or projects funder for replication and future usages.Publish papers on top-tier conferences/journal in the relevant fields.Collaborate with engineering team to develop demonstration systems.Managing teams computing resources to support teams research experiments.Job Requirements:PhD or Masters Degree in computer science/engineering or related fields.Experience with large language model, speech and text data processing, data preparation and organization for deep learning model training.Strong in the following programming languages: Python, Pytorch, C/C++, Linux Bash/Shell.Having prior experience with speech processing frameworks such as ESPnet, Wenet, Whisper, Kaldi.Having solid background in deep learning for speech and language technologies, transformers and end-to-end ASR.Prior experience with ASR for accented English and/or children speech is strongly preferred.Having publications at top-tier conferences in speech processing fields such as Interspeech, ICASSP is a plus.We regret that only shortlisted candidates will be notified.Hiring Institution: NTU

Nanyang Technological University