For my project, I would like to explore phonetic differences between the accent's of native and non-native speakers and how the rate of adaption of phonetic features for the English language compares between the two groups. Specifically, I would like to examine the frequencies and occurrence count of these phonemes between the speakers and see if some of them adopt at higher rates than others.
I plan to use the Speech Accent Archive dataset on Kaggle, which has mp3 audio snippets for non-native English speakers and the respective biographical relevant information. I will most likely need to do some preprocessing and clean up before encoding. For the encoding I will use a standard Hidden Markov Model to encode the phonemes used in the audio files, then create a phonetic data frame or CSV to analyze later.
In term's of analysis, I will most likely compare and examine the results of the produced phonetic data frame to the biographical information of the speakers as well as analyze the proximity of their native language to English to explain outliers or significant differences in accent.
The presentation will talk about how I designed the data processing, the encoding, the analysis, as well as any significant factors or limiting variables that were not considered in the project due to time constraints or complexity.