Translation into the speech of unwritten languages

We are building a system for automatic translation of English into languages that lack a writing system, so that the translation has to generate speech directly, without going through text. The focus is on translation of the Bible.

We have a working prototype, but need a student to package the system in a way that makes it easily usable by anyone. Additional research topics may include: evaluating the system against alternative approaches; developing new evaluation metrics; developing methods for tokenization of speech.

 

Name of research group, project, or lab
NLP Group
Why join this research group or lab?

The NLP group more broadly investigates computer processing of less-represented human languages, and the theory of language models.

Logistics Information:
Project categories
Computer Science & Engineering
Student ranks applicable
First Year
Sophomore
Junior
Senior
Graduate Student
Student qualifications

Experience with PyTorch and HuggingFace is a must. Enthusiasm for languages and/or the Bible is a plus!

Hours per week
3 credits / 12+ hours
Compensation
Research for Credit
Paid - General
Number of openings
1
Project start
As soon as possible
Contact Information:
Mentor
dchiang@nd.edu
Associate Professor
Name of project director or principal investigator
David Chiang
Email address of project director or principal investigator
dchiang@nd.edu
1 sp. | 0 appl.
Hours per week
3 credits / 12+ hours
Project categories
Computer Science & Engineering