Postgraduate Course: Speech Synthesis (LASC11062)
Course Outline
| School | School of Philosophy, Psychology and Language Sciences |
College | College of Arts, Humanities and Social Sciences |
| Credit level (Normal year taken) | SCQF Level 11 (Postgraduate) |
Availability | Available to all students |
| SCQF Credits | 10 |
ECTS Credits | 5 |
| Summary | This course covers the current state-of-the-art in speech synthesis. The course starts with the historical context, so that students understand how we arrived at the state-of-the-art, and concludes with the most recent neural-network systems. The course also provides some foundation material on machine learning and neural networks, and coverage of both classical and machine-learning-based speech signal processing, as used for speech coding and synthesis. The course may also touch on issues surrounding speech synthesis, such as its use in creating deepfakes. |
| Course description |
This course is delivered using a variety of teaching and learning methods. The early part of the course uses a flipped classroom in which students watch videos and do readings, which then form the basis of the interactive weekly class. The later part of the course involves reading recent research papers, which are then elaborated and discussed in class. A single practical assignment, performed in weekly supervised computer lab sessions, forms an integral part of the course. In the assignment, students record themselves in a professional recording studio, then use their recordings to build a neural speech synthesiser, which they subsequently evaluate using a listening test. The coursework is written up in the style of journal paper.
|
Information for Visiting Students
| Pre-requisites | None |
| High Demand Course? |
Yes |
Course Delivery Information
|
| Academic year 2026/27, Available to all students (SV1)
|
Quota: 0 |
| Course Start |
Semester 2 |
Timetable |
Timetable |
| Learning and Teaching activities (Further Info) |
Total Hours:
100
(
Lecture Hours 18,
Supervised Practical/Workshop/Studio Hours 18,
Programme Level Learning and Teaching Hours 2,
Directed Learning and Independent Learning Hours
62 )
|
| Assessment (Further Info) |
Written Exam
0 %,
Coursework
100 %,
Practical Exam
0 %
|
| Additional Information (Assessment) |
4000-word lab report worth 100% |
| Feedback |
Class-wide formative feedback will be provided based on work previously submitted for Speech Processing (from multiple students, anonymised), since the style and format are similar. This feedback takes the form of videos, slides, and blog posts, with follow-up questions answered via the course forum, in class, and in lab sessions.
Comments will be provided on submitted coursework. A structured marking scheme will be used. All students will have the opportunity for an individual 15-minute summative feedback session with the course organiser, after it is returned.
|
| No Exam Information |
Learning Outcomes
On completion of this course, the student will be able to:
- understand speech synthesis methods currently in use, and the historical developments that underpin them
- be familiar with speech signal processing and coding techniques that are used for speech synthesis
- have the practical experience of building a synthetic voice
- be able to discuss current issues in speech synthesis and be well-placed to understand future developments
- have improved their scientific written communication skills
|
Reading List
Indicative reading list:
Paul Taylor ¿Text-to-speech synthesis¿, 2009, Cambridge University Press, Cambridge
Chengyi Wang et al. ¿Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers¿ DOI: 10.48550/arXiv.2301.02111
Latest information is on the course webpage: https://speech.zone/courses/speech-synthesis
|
Additional Information
| Graduate Attributes and Skills |
Critical thinking; Problem solving; Curiosity; Collaboration; Communication; Reflection; Inclusivity; Adaptivity; Data and Digital Literacy; Individuality
The course materials require students to take advantage of the wide variety of learning opportunities provided: videos, textbook readings, interactive classes, traditional lecture-style teaching, research papers, and practical laboratory work. Interactive classes involve some in-class group work. In the practical assignment, students work in pairs to record speech in a professional recording studio and then work individually to use their recordings to build a synthetic voice. Students will need to master advanced computational skills (e.g., remote access to a compute cluster, GPU computation), which they will be taught in the lab. The synthetic voice is evaluated using a hypothesis-driven listening test. Good time and workload management are required to balance the elements of the course and spend an appropriate amount of effort on each activity. The coursework instructions are deliberately somewhat under-specified, to encourage students to develop their individual project planning, organisation, and execution skills. The assignment is written up in the style of a journal paper.
Core skills gained or developed on this course: Critical thinking, analysis, and evaluation; Data collection and analysis; Enhanced programming / coding skills; Independence; Preparation, planning and organisation; Problem solving; Academic reading skills; Report writing; Research skills; Resilience; Time management; Workload management; Written communication; Writing clearly and concisely; Critical reading of recent research publications; Scientific writing, following a journal style guide; Designing and implementing an original algorithm; Using a professional recording studio; Experimental design, including the use of human subjects for perceptual testing. |
| Keywords | speech synthesis,machine learning,neural networks,signal processing |
Contacts
| Course organiser | Prof Simon King
Tel: (0131 6)51 1725
Email: Simon.King@ed.ac.uk |
Course secretary | Ms Sasha Wood
Tel:
Email: swood310@ed.ac.uk |
|
|
University Homepage