Ross McDonald
MSc Speech and Language Processing Graduate
Freelance Developer
MSc Speech and Language Processing Graduate
Freelance Developer
My dissertation was completed in conjuction with Cisco-WebEx. The goal of it was to automatically classify a speaker's characteristics in order to check the variation of speakers within a dataset. I chose to focus my work on classifying speakers' accents, rather than age or gender for which many successful models already exist.
I built models using traditional machine learning methods (k-NNs, SVMs and RFCs) and neural networks (CNNs and ResNet). I used the different methods as I wanted to compare the difference in accuracy classification. My final results were very promising, I was able to achieve a 92%+ accuracy on a binary classification using a CNN. The dataset I was using, Mozilla Common Voice Dataset, had issues with quality and label accuracy, which reduced my model accuracy.
The code for my dissertation is available on my GitHub
I compeleted my masters in September 2022 at the University of Edinburgh. This course is interdisciplinary, combining computer science, lingusitics, engineering and psychology. The topics I studied included Machine Learning, Reinforcement Learning, NLP, Speech Recognition, Data Engineering, Data Science, Data Analysis and UX.
A large part of the course focused on practical application of the content via assignments and reports.
You can read more about the programme here
Since graduating I have been using Kaggle to develop my data science and analysis skills. I wanted to continue working with CNNs, having used them in my dissertation, and so I have worked with two different datasets of animal images to build classifiers. In one which contained three different classes of animal faces I achieved an accuracy of 96%+. The other contained ten classes, and included different angles of the animals, for which I have produced an accuracy of 75%+ so far. I am looking to improve this and use a third dataset to develop an object detection model.
I recieved an honourable mention for my submission to the Kaggle Survey 2022 in which I focused on representation and professional variety within the industry. You can read this here.
I began my programming career as a Web Developer, having worked for a marketing agency and also freelance. I have worked with a range of clients, such as the BBC, Tesco Bank and online stores.
An example of my work can be found here, it is a commission for a local artist's portfolio.
I also created a top trumps Pokemon game using the Pokeapi which you can play here
I am passionate about accessibility and diversity within the tech industry and I have volunteered with Code Your Future and Code First Girls. Both of these two organisations work to help improve accessibility and support for groups which are minorities within the tech sector.
I am now a freelance tutor with Code First Girls, teaching several classes a week.
You can reach me at rosspmcd@gmail.com