Response to Speech Synthesis feedback of 2026-02

[Simon King]

Thank you for the useful feedback (from 22 responses), regarding the Speech Synthesis course. I have clustered your comments, and my responses are below.

I have briefly summarised the most frequently mentioned “Keep doing this” points without adding any response, then responded to each “Change this” point made by at least 2 people. All feedback was carefully considered and even points made by just one person are still taken into account.

Please do feel free to follow up and offer more feedback or clarification.

[Simon King]

Keep doing this

Number of people mentioning each point is given in parentheses.

In-class interactivity (12)

speech.zone content, structure, etc (8)

Videos (8)

Division into classes and lab sessions (4)

The lab tutor (2)

Quizzes on speech.zone (2)

[Simon King]

Too much video content

Yes, I accept there are a lot of videos in the first half of the course. It is rather difficult to reduce the quantity since dropping entire videos doesn’t always make sense. For next year, I will think carefully about a better way to reduce the quantity whilst keeping the course coherent.

Please update me at the end of the course, given that there are not many more videos to watch.

For this year, I have taken the following action:

• Module 7 is non-examinable. If you need to skip some videos, skip these.

[Simon King]

Too much reading

Do please remember that only the essential readings are required to pass the course. Everything else is optional, and only needed if you are aiming for a higher mark.

For this year, I will take the following action:

• Limit the number of essential readings in the remaining modules
• Reduce the number of recommended readings, making them extra to clarify that you only need to read (some of) them if you are particularly interested in the topic. For example, Module 8 has only one essential reading and the rest are extra

[Simon King]

Undergraduate students are in a minority

Yes, that’s always true in this course. You mustn’t let this overwhelm or intimidate you because, in my long experience, undergraduate students usually do extremely well on this course.

Some comments implied that undergraduate students had to do more “catching up” for the course. I don’t think that’s necessarily true – Speech Processing provided the best possible preparation. You also have twice as much time available to spend on this course (it’s 20 credits for UG and only 10 credits for PG).

It is inevitable that most postgraduate students are taking closely-related courses in Natural Language Processing and Machine Learning. In class, I make connections to those subjects, to help those students. But this doesn’t imply that undergraduates need also learn that material.

[Simon King]

Can you offer office hours or “remedial” sessions?

Yes, I can! I already provided 3 such sessions in Week 4, although relatively few students took advantage of these. I will offer more during the second half of the course.

In these sessions, you can ask about any and all course content, or indeed about background material from Speech Processing.

[Simon King]

Miscellaneous points

Not necessarily made by 2+ people, but I will respond to some anyway

Trouble hearing Simon in class: please talk to me in the next class to help me understand whether this is me, or the room audio system.

No camera in classroom, so recordings are voice+screen only: unfortunately, this is a limitation of the room we are using, so I can’t fix this. Remember that recordings are intended for revision only and not a replacement for attending class (unless you are unwell).

Video speed buttons only offer 3 levels: already solved – see this forum topic

Button to jump to next video does not work (jumps to readings as well as videos): request noted

Labs should involve more coding: this course does not have a coding pre-requisite, to keep it accessible to a wide range of students. You can still improve your coding by reading the source code for the model we are using, if you wish, or by implementing a text-selection algorithm.

Assignment needs a 1-page version: I have added the option to “Click here to view as a single page”. Warning: the lab instructions may be updated and refined later, so you should use the latest website version and not make a copy.

[Author]

Posts