Robert Fromont

Annotations of various types can be automatically generated or manually added.

The transcripts and annotations can be searched for particular text or regular expressions. The search results, or entire transcripts, can be viewed or saved in a variety of formats, and the related parts of the recordings can be played or opened in acoustic analysis software, all directly through the web-browser.

Combining the signal data, the raw orthographic transcriptions, and some third-party data and tools, the transcripts can be automatically annotated, for example:

• Token tagging (pronuncation, part of speech, lemma, frequency)
• Syntactic parsing
• Forced Alignment (determine the start/end time of each individual speech sound)
• Statistical tagging (frequency, type/token ratio, speech rate, LIWC)

Transcripts and annotations can be searched, analysed, and data extracted is a variety of formats, using the browser interface, or programmatically using client packages for R, Python, JavaScript, and Java.

Users of LaBB-CAT include:

• the NZILBB at the University of Canterbury (Jennifer Hay, Margaret Maclagan, Jeanette King, Kevin Watson, Lynn Clark, Megan McAuliffe),
• the Glasgow University Laboratory of Phonetics (Jane Stuart-Smith),
• Arizona State University (Visar Berisha, Julie Liss),
• ZAS Berlin (Stefanie Jannedy),
• University of HawaiÊ»i at Mānoa (Katie Drager),
• Adam Mickiewicz University in Poznań (Kamil Kaźmierski),
• Medical University of South Carolina (Boyd Davis, Charlene Pope),
• University of Oxford (Sarah Ogilvie),
• Griffith University (Gerry Docherty), and
• Australian National University (Ksenia Gnevsheva, Catherine Travis).

A cross-platform app for speech elicitation tasks. Currently integrates with LaBB-CAT, where you can define a 'task' as a series of 'steps' that display prompts and text for the participant to read aloud, or ask meta-data questions.

Features include:

• OS or Android mobile app, and/or Chrome/Firefox web-app
• display paginated textual instructions and prompts
• ask questions; answers can be typed text, numbers, checkboxes, selection from a list, dates/times
• arbitrary validation of input is possible (e.g check that one date is after another, etc.)
• steps can be shown contingent on the answers to previous questions
• record participant's speech
• upload data to a LaBB-CAT corpus
• stimuli can be text, images, or video
• stimuli can be randomly sampled from a list

ElicitSpeech has been used by:

• the NZILBB at the University of Canterbury (Sidney Wong, Gia Hurring),
• Aural Analytics (Visar Berisha and Julie Liss)
• Mayo Clinic Arizona (Catherine Chong)

A noisy recording of three digits is played, and the participant must enter the digits, before moving to the next recording. As the test proceeds, the digits become harder to hear.

An implementation for New Zealand English has been published at nzhst.canterbury.ac.nz but the system can be easily ported to other languages. Versions also exist for Te Reo Māori and Malay.

The participant hears noisy recording of a sentence and presses buttons in a matrix to identify what was said.

The stimulus can be audio only, or audio and video of the speaker's face as they speak.

Modules include:

• freeform rich-text pages
• contact forms
• blogs/newsletters
• catalogs
• file downloads
• events calendar
• image galleries
• ...and more

Sites using Hexagon include:

• carolinaconversations.musc.edu
• robert.fromont.nz
• www.fromont.net.nz
• www.ceciliaborghi.com