SOAS University of London

SOAS World Languages Institute

Persephone: a tool for the automatic phonemic transcription of endangered languages

Prof Christopher Cox (Carleton University), Dr Hilaria Cruz (University of Kentucky) and Dr Alexis Michaud (CNRS)

Date: 12 November 2019Time: 2:00 PM

Finishes: 13 November 2019Time: 5:00 PM

Venue: Russell Square: College Buildings Room: L67

Type of Event: Workshop

Asia Beyond Boundaries (ERC) and the Endangered Languages Archive (ELAR) present:

Persephone: a tool for the automatic phonemic transcription of endangered languages, Tuesday 12 7 Wednesday 13 November 2019, 2-5pm. 

With speakers:

  • Prof Christopher Cox (Carleton University)
  • Dr Hilaria Cruz (University of Kentucky)
  • Dr Alexis Michaud (CNRS)

Many of the world's languages are unstudied and on the verge of disappearance. A few linguists strive to address this problem, but compared to Indo-European languages, there is much more work to do, with only a fraction of the manpower and under immense time pressure. The automatic phonetic transcription of spoken languages using neural networks, provides a ray of hope that the drugergy of meticulous transcription of audio files can be automated and thus, much larger collections of more endangered languages will become easier to compile.

This workshop introduces Persephone (/pərˈsɛfəni/), an automatic phoneme transcription tool. As opposed to traditional speech recognition tools Persephone is designed for situations where training data is limited, perhaps as little as an hour of transcribed speech. Such limitations on data are common in the documentation of low-resource languages. It is possible to use such small amounts of data to train a transcription model that can help aid transcription, yet such technology has not been widely adopted.

The event includes three presentations that showcase the use of Persephone for Sino-Tibetan and Athabaskan languages, and there will also be a hands-on training the following day for linguistics who want to try Persephone on their own data.

Please sign up via Eventbrite as places are limited!

Contact email: