Projects

Yoruba Speech Dataset: A dataset for low-resource speech tasks.

• 3 hours of audio recordings from a native yoruba speaker.
• Can be used for low-resource speech tasks
• Details of the project can be found on my github repo

Exploring Transfer of Self-supervised English Pretrained Model to Yoruba and French Languages in Very Low-Resource Settings

• We use a self-supervised pretrained model trained on large amount of English speech
• Model is finetuned on 1 hour of yoruba and french audio with labels
• We compare this to training on 1 hour of audio in the target language from scratch.
• Audio samples are from only 1 speaker
• Surprisingly, we got better results finetuning on a single speaker for Yoruba language. Many factors could have contributed to this including gender (Male for Yoruba and Female for French, quality of audio recording etc)
• Details of the project can be found on my github repo

The Matching Game

• A game written in React/Javascript to test attention and retention
• Game hosted on netlify. You can play using the game link