This course provides an overview of the theoretical and practical issues related to automatic speech recognition, a rapidly advancing research field bridging AI, linguistics and computational modeling.
From a theoretical perspective, students will be able to understand and explain important aspects of ASR such as the feature extraction, acoustic modeling (using Deep Neural Networks or Gaussian mixture models), training and decoding algorithms (Baum-Welch, Viterbi), the pronunciation vocabulary, language modeling, keyword spotting, evaluation procedures, and scalability of the approach.
From a practical point of view, students will familiarize themselves with speech decoding techniques via existing ASR software, web interfaces or Deep Neural Networks, so that they are able to perform mid-scale ASR experiments. A number of default experiments are pre-specified, but students are encouraged to explore their own research questions, e.g., the discovery of structure in speech, the use of prediction in speech decoding, accent detection, word search, entropy, or the relation between speech and artificial intelligence. The focus is on gaining insight in speech decoding, not on development of implementations.
In principle, each students writes his/her own scientific report, but cooperation may apply - depending on group size. |
 |
|
|
In the first part of the course, relevant aspects of speech production, speech perception, and acoustics are addressed. Thereafter, we discuss the theoretical and practical aspects of Automatic Speech Recognition.
Speech production is a process in which some spoken message (e.g. a sequence of words or word-like units) is transformed into a continuous speech signal (spoken utterance). Due to the personal anatomical characteristics in a speaker’s vocal tract, speech signals that represent the same message usually show a tremendous amount of variation.
Next, we address the question how the reverse process might take place, i.e., the decoding of a spoken utterance in terms of discrete elements (e.g., words). In particular, we will explore various approaches for Automatic Speech Recognition (ASR), including the hidden Markov models (HMMs) based ASR, and the artificial net based approaches in ASR.
Lectures are given to provide students with background knowledge on ASR. Students will define their own research questions which can be answered by means of ASR technology. Students meet on a regular basis to present their progress to fellow students and to discuss any challenges encountered. A scientific report is written on the research carried out. Depending on the topics chosen, this report is individual or team-work.
|
|
|
|
Knowledge of statistics and scripting/programming.
Knowledge of AI and or deep learning and or data science will be helpful.
|
|
|
Assessment will take place on the basis of an individual thesis. This thesis is either based on own experiments or on experiments carried out in a team.
|
|
|