Title: "Toolkit for acoustic-phonetic analysis of naturalistic speech data"

Speaker: Ethan Kutlu, PhD

One of the major limitations in the speech sciences is access to naturalistic input in experimental setups and the difficulty of translating in-lab experimental setups to real-world situations. Many scholars studying speech production or perception rely on in-lab participants and audio recordings, which can introduce psychological effects on outcomes and restrict the sociolinguistic contexts examined. The Toolkit for Acoustic-Phonetic Analysis (TAPA) is a collection of various modules that can be used independently or combined to automatically acquire, process, and analyze naturalistic, multi-speaker acoustic-phonetic data. Currently, the toolkit's main components are the detection, transcription, and analysis of fundamental frequency and vowel formants in multi-speaker conversations. This talk demonstrates the application of this toolkit to spontaneous multi-party. The toolkit is fully open-source and does not require advanced computational knowledge. The scripts can be expanded to accommodate additional acoustic-phonetic features based on researchers' needs.

Attendees are encouraged to join in person in HSAB 2004. Seating is limited (first-come, first-served, ~50 seats), but the seminar will also be available via Zoom.