Emu Workshop

Emu is a speech database management system:

• Developed by Raphael Winkelmann at IPS in Munich.
• Developed out of The Emu Speech Database developed at Macqurie Uni by Steve Cassidy and Jonathan Harrington and others.

• Fast and powerful way to process speech files particularly if they are part of large corpora.
• Hierarchical analysis
• Access to statistical tools in R

• How to visualise speech files in Emu The Emu Webapp
• Formats for your data
• Best practise for naming files
• How to setup a database in EmuR
• some basic visualisation in R

• Many of you will be familiar with Praat.
• What does Emu do that Praat does not?
• Platform Independent/Browser Based (works best with Chrome/Firefox)
• Interfaces easily with R

• Navigate to: http://ips-lmu.github.io/EMU-webApp/
• Drag and Drop a pair of files (wav and textgrid)
• These can be up to an hour long depending on quality of audio
  • (there is a limit of 2Gb per tab in Chrome)

• See help in the Emu Web App
• Full set of Keyboard shortcuts
• Feature requests welcomed: Link

• Spectrogram
• Download Textgrid
• Download JSON

• No access to speech analysis tools
• No direct access to statistical and visualisation tools within R

• To install R - go to: https://cran.ms.unimelb.edu.au
  • choose your platform (Windows, OSX, Linux)
• To install Rstudio go to: Download Rstudio
  • choose your platform (Windows, OSX, Linux)

Once R and Rstudio are installed:

install.packages("emuR")

To load the library, use:

library("emuR")

• Create and load a demo emuDB

create_emuRdemoData()

followed by

ae = load_emuDB(file.path(tempdir(), "emuR_demoData", "ae_emuDB"))

• Query the emuDB for all /n/ segments of the “Phonetic” annotation level

    sl = query(ae, "Phonetic == n")

• Extract the according formant values -

td = get_trackdata(ae, sl, onTheFlyFunctionName = "forest")

Serve emuDB to EMU-webApp for annotation / visual inspection -

serve(ae)

For a more explicit introduction see

vignette("emuR_intro")

TBA