Postgraduate Course: Automatic Speech Recognition (INFR11033)
Course Outline
| School | School of Informatics |
College | College of Science and Engineering |
| Credit level (Normal year taken) | SCQF Level 11 (Year 4 Undergraduate) |
Availability | Available to all students |
| SCQF Credits | 10 |
ECTS Credits | 5 |
| Summary | This course covers the theory and practice of automatic speech recognition (ASR), with a focus on the statistical approaches that comprise the state of the art. The course introduces the overall framework for speech recognition, including speech signal analysis, acoustic modelling using hidden Markov models, language modelling and recognition search. Advanced topics covered will include speaker adaptation, robust speech recognition and speaker identification. The practical side of the course will involve the development of a speech recognition system using a speech recognition software toolkit. |
| Course description |
* Signal analysis for ASR
* Statistical pattern recognition (Bayes decision theory, Learning algorithms, Evaluation methods, Gaussian mixture model, and EM algorithm)
* Hidden Markov Models (HMM)
* Context-dependent models
* Discriminative training
* Language models for LVCSR (large vocabulary continuous speech recognition)
* Decoding
* Robust ASR (Robust features Noise reduction, Microphone arrays)
* Adaptation (Noise adaptation, Speaker adaptation/normalization, Language model adaptation)
* Speaker recognition
* History of speech recognition
* Advanced topics (Using prosody for ASR, Audio-visual ASR, Indexing, Bayesian network)
* speech recognition applications (including privacy implications)
Relevant QAA Computing Curriculum Sections: Artificial Intelligence, Natural Language Computing
|
Entry Requirements (not applicable to Visiting Students)
| Pre-requisites |
It is RECOMMENDED that students have passed
Speech Processing (LASC11065)
|
Co-requisites | |
| Prohibited Combinations | |
Other requirements | This course is open to all Informatics students including those on joint degrees. For external students where this course is not listed in your DPT, please seek special permission from the course organiser.
Some general mathematical ability (at pre-honours Maths for Informatics level) is essential; Special functions log, exp are fundamental; mathematical notation (such as sums) used throughout; some calculus. Probability theory is used extensively: joint and conditional probabilities, Gaussian and multinomial distributions.
Programming using Python or shell scripting is required for the practicals and coursework. |
Information for Visiting Students
| Pre-requisites | None |
| High Demand Course? |
Yes |
Course Delivery Information
|
| Academic year 2017/18, Available to all students (SV1)
|
Quota: None |
| Course Start |
Semester 2 |
Timetable |
Timetable |
| Learning and Teaching activities (Further Info) |
Total Hours:
100
(
Lecture Hours 15,
Seminar/Tutorial Hours 8,
Supervised Practical/Workshop/Studio Hours 5,
Feedback/Feedforward Hours 6,
Summative Assessment Hours 2,
Programme Level Learning and Teaching Hours 2,
Directed Learning and Independent Learning Hours
62 )
|
| Assessment (Further Info) |
Written Exam
70 %,
Coursework
30 %,
Practical Exam
0 %
|
| Additional Information (Assessment) |
Assessed coursework will comprise the development of a speech recognition system using a standard software toolkit.
You should expect to spend approximately 30 hours on the coursework for this course.
If delivered in semester 1, this course will have an option for semester 1 only visiting undergraduate students, providing assessment prior to the end of the calendar year. |
| Feedback |
Not entered |
| Exam Information |
| Exam Diet |
Paper Name |
Hours & Minutes |
|
| Main Exam Diet S2 (April/May) | | 2:00 | |
Learning Outcomes
On completion of this course, the student will be able to:
- Describe the statistical framework used for automatic speech recognition.
- Understand the weakness of the simplified speech recognition systems and demonstrate knowledge of more advanced methods to overcome these problems.
- Describe speech recognition as an optimization problem in probabilistic terms.
- Relate individual terms in the mathematical framework for speech recognition to particular modules of the system.
- Build a large vocabulary continuous speech recognition system, using a standard software toolkit.
|
Reading List
* John N. Holmes, Wendy J. Holmes, "Speech Synthesis and Recognition", Taylor & Francis (2001), 2nd edition
* Xuedong Huang, Alex Acero and Hsiao-Wuen Hon, "Spoken language processing: a guide to theory, algorithm, and system development", Prentice Hall (2001).
* Lawrence R. Rabiner and Biing-Hwang Juang, "Fundamental of Speech Recognition", Prentice Hall (1993).
* B. Gold, N. Morgan, "Speech and Audio Signal Processing: Processing and Perception of Speech and Music", John Wiley and Sons (1999).
|
Contacts
| Course organiser | Prof Stephen Renals
Tel: (0131 6)50 4589
Email: |
Course secretary | Mr Gregor Hall
Tel: (0131 6)50 5194
Email: |
|
© Copyright 2017 The University of Edinburgh - 6 February 2017 8:09 pm
|
University Homepage