MSHLT Student Laura Montenegro

See this post on LinkedIn

Our Master’s in Human Language Technology program had one other student graduate in the fall semester: Laura Montenegro. For her internship with Cobalt Speech, Laura evaluated a hybrid ASR model (using a Hidden Markov Model-Deep Neural Network) for low-resource languages using Cobalt’s existing tools.

Although “low-resource language” is the standard term to refer to languages without large existing datasets, I find the term “data-scarce language” to be less prone to misunderstanding; the “resource” that is low is prepared datasets, not necessarily money or people.

By definition, it’s hard to find datasets for languages that are considered “low-resource,” but Laura found a workable amount of audio data for the Ewe language on HuggingFace—great practical experience for future work in NLP and speech technology!

Wishing you all the best in your future endeavors, Laura!