Tutorial day
Monday 2nd December
The first day of SST2024 will be the tutorial day, held in the same venue as the main conference. Tutorials are intensive instructional sessions that provide a comprehensive introduction to established or emerging research topics of interest to the SST community. Four tutorials will be conducted on the scheduled day, with two running in parallel sessions in the morning and two in the afternoon (see below). Each session is 3 hours in duration with a break. Participants can choose to attend two out of the four sessions (i.e. one of two tutorials available at each time slot).
Participants have the option to register for the tutorial day either independently, without attending the main conference, or as an addition to their main conference registration. Lunch and morning and afternoon tea are included in the in-person registration fee.
After registration, please take a moment to indicate which two tutorial sessions you plan to attend via completing this short survey.
Tutorial 1: Phonetic documentation: How to develop a comprehensive description of a language’s sound system
Rosey Billington (Australian National University)
Abstract: Phonetic studies typically focus on highly specific speech phenomena. For some languages, such studies form part of an extensive body of research conducted over many decades, by many different researchers, cumulatively building a detailed understanding of the phonetic structures of the language. For minority and low-resource languages, such studies may be the only phonetic research on the language, and for the majority of the world’s languages, there is no phonetic research at all. This tutorial will focus on phonetic documentation, here referring to the systematic collection and analysis of speech data, particularly focusing on the phonetic realisation of phonological contrasts in a given language. We will discuss motivations for phonetic documentation, what comprehensive phonetic documentation might look like, and whether this is even possible. We will then consider key aspects of planning the collection or collation of acoustic phonetic data and undertaking appropriate analyses, and considerations for managing data and sharing findings.
Target audience: This tutorial is open to researchers at any level, ideally with some knowledge of at least undergraduate-level phonetics and phonology. It will be of particular interest to participants considering or engaged in research on understudied languages, either through collecting new data or working with existing materials, and participants who are broadly interested in phonetic typology.
Requirements: Participants should bring headphones and a laptop with Praat installed. No preparation will be required before the tutorial. Example audio files and suggested readings will be shared during the session.
Tutorial 2: An introduction to methods for acoustic analysis of voice quality
Joshua Penney & Hannah White (Macquarie University)
Abstract: In this tutorial, we will introduce a selection of methods for examining voice quality acoustically. The first session will provide a brief overview of core theoretical concepts and introduce acoustic measures that correspond to particular voice qualities, focussing on differences between modal, creaky, and breathy voices. We will demonstrate how these measures can be estimated, examined, and visualised. The second session will focus specifically on creaky voice, introducing semi-automatic methods for identifying creaky portions of speech. The tutorial will include opportunities for hands on practical activities. Participants are also welcome to bring examples of their own data to work with (ideally limited to a few audio files to constrain processing time).
Target audience: This tutorial is introductory and aimed at students/researchers with no (or minimal) experience in acoustic analysis of voice quality, although some basic phonetics knowledge is assumed. Participants should have basic competence working in R.
Requirements: The tutorial will include practical activities enabling participants to work with sample voice quality data. Participants should bring a laptop with R/RStudio and the ggplot2, tidyverse, ands modes (https://cran.r-project.org/src/contrib/Archive/modes/) packages installed.
The methods that will be demonstrated in this tutorial make use of specialised software, some of which require a Matlab license. Participants wishing to follow along and/or apply these methods to their own data will need to install the software listed below:
- VoiceSauce (requires Matlab): https://phonetics.ucla.edu/voicesauce/
- Reaper (requires installation and compilation): https://github.com/google/REAPER
- MacReaper (available for Mac only): https://kjdallaston.com/projects/
Note that access to this software is not necessary to take part in the session. All methods will be demonstrated visually, and example data will be provided for all practical activities.
Tutorial 3: Cochlear implant sound processing and perception
Brett A. Swanson (Cochlear Ltd)
Abstract: A cochlear implant restores a sense of hearing to a person with severe to profound deafness. The objective of this tutorial is to provide an overview of signal processing and sound perception for cochlear implants, with emphasis on speech understanding in noisy conditions. Multiple microphone beam-forming techniques can provide large improvements in speech intelligibility when speech and nose are spatially separated. When the speech and noise come from the same direction, noise reduction algorithms seek to identify and exploit specific noise characteristics. Many noise reduction algorithms can be understood as forms of multi-band AGC. A simple model that calculates how the signal-to-noise ratio is modified by the applied gains can make useful predictions of speech intelligibility under a variety of processing algorithms. These algorithms are commonly used in conjunction with auditory scene classification.
Requirements: The tutorial will make use of Nucleus Toolbox for MATLAB, which has recently been released as open-source software (https://github.com/Cochlear-Research/nucleus-toolbox). It implements the core sound processing algorithm used in the Nucleus cochlear implant system manufactured by Cochlear Ltd.
Tutorial 4: Prosodic prominence – a cross-linguistic exploration
Sasha Calhoun (Victoria University of Wellington)
Abstract: Across many languages, prosodic prominence has a key function in highlighting or emphasising a part of an utterance, e.g. because it is focal, contrastive or important, and is therefore a core part of the prosodic system of most languages. Prosodic research over the past decades has established that there is substantial variation across languages in how prominence is expressed, and that prominence must be analysed as part of the phonology of the language. Nonetheless, it is still common to see the assumption that prominence can be directly measured using acoustic cues like mean F0, duration and intensity. We explore how this can lead to erroneous conclusions, and better approaches, through examples of the three main prosodic prominence types: stress-based, phrase-based and register-based (Kügler & Calhoun, 2020).
Target audience: The tutorial is open to researchers at any level, but is designed for those with little or no previous experience of prosodic analysis. Participants should ideally have some background in phonetics, but those working in related areas are most welcome.
Requirements: Participants need to bring a laptop with headphones and Praat installed. Sound files and Praat TextGrids will be sent to tutorial participants in advance. If you have sound files of a language you would like to discuss, please send these to Sasha well in advance. Note in this case you must have permission to share these files with other tutorial participants.
