Speech Technology Magazine


Aculab Collaborates with the University of York for Language and Linguistic Science Project

Applicants sought for 2019 WRoCAH AHRC Collaborative Doctoral Award Studentships.
Posted Nov 8, 2018
Page1 of 1
Bookmark and Share

Aculab, a global speech technology and development API provider, is pleased to announce its collaboration with the University of York in a WRoCAH funded language and linguistic science research project entitled Towards linguistically-informed automatic speaker recognition.

Automatic speaker recognition (ASR) is increasingly used by commercial institutions to identify individuals from their voices, and by forensic labs for voice evidence in legal cases. Despite their very low error rates in experiments, little is known about what information is actually captured by state-of-the art ASR systems. This collaborative project, involving Aculab and the University of York, will examine how speaker-characterizing information captured by ASR systems maps on to linguistic properties of the voice. This study has important implications for the development and improvement of ASR, enhancing public and legal/judicial understanding of ASR, and improving voice analysis in forensic cases.

Understanding and modeling the human voice in all its complexity is a key issue in both the humanities (linguistics, phonetics) and the sciences (engineering, computer science). Different disciplines approach this problem in fundamentally different ways. Despite the clear overlap in interests, very little work has been situated at the intersection between linguistics and speech technology.

The project will contribute to the small but emerging body of research investigating how linguistic information can help us better understand ASR systems. The project will address three key questions:

  1. To what extent do ASR systems capture tangible linguistic properties of a voice?
  2. By understanding what information is captured by ASR systems, can we predict which speakers will be problematic for the system?
  3. Can linguistic information be used to improve the performance of ASR?

Page1 of 1