neulab/cmulab_elan_extension

CMU Linguistic Annotation Backend plugin for ELAN

CMU Linguistic Annotation Backend plugin for ELAN

This plugin is still a work in progress. Currently it allows users to do automatic speaker diarization and phone transcription using the CMU Linguistic Annotation Backend server APIs. Users can even upload data to fine-tune the pre-trained phone recognizer (Allosaurus)

Setup

Note: The plugin requires Python 3.

Linux

1. Download the latest version of ELAN from here and install it:

wget https://www.mpi.nl/tools/elan/ELAN-XX_linux.tar.gz
tar xzf ELAN-XX_linux.tar.gz

2. Download a copy of this repo and unzip it. Copy the cmulab_elan_extension-main/ folder into ELAN's extensions dir (ELAN-XX/lib/app/extensions/).

Mac

1. If ELAN is not already installed on your Mac, download the latest .dmg installer and install it. It should be installed in the /Applications/ELAN_XX directory, where XX is the name of the version.
2. Download this zip file and unzip it. You should see a folder named cmulab_elan_extension-main containing the contents of this repo.
3. Right-click ELAN_XX and click "Show Package Contents", then copy your cmulab_elan_extension-main folder into ELAN_XX.app/Contents/app/extensions.

Note: The built-in Tk GUI library in Apple-supplied Python 3 in some macOS versions (such as macOS 12 Monterey) have bugs that might cause dialog boxes to not display properly. In that case please install the latest python from python.org.

Windows

1. Download the latest version of ELAN from here and install it.
2. Download a copy of this repo and unzip it. Copy the cmulab_elan_extension-main/ folder into ELAN's extensions dir (ELAN-XX/app/extensions/).
3. Install Python 3 if it isn't already installed.

Troubleshooting

If the extension is not visible in the "Recognizers" tab or doesn't run after following the above instructions, please make sure that the contents of the zip file are extracted to:

ELAN-6-2/app/extensions/cmulab_elan_extension-main/ rather than:

ELAN-6-2/app/extensions/cmulab_elan_extension-main/cmulab_elan_extension-main/

Also the downloaded cmulab_elan_extension-main.zip should be removed from the extensions directory ELAN-6-2/app/extensions/, otherwise ELAN tries to load the zip file rather than the extracted cmulab_elan_extension-main/ folder.

Instructions

Start ELAN with the provided test audio file

ELAN_6-3/bin/ELAN allosaurus-elan/test/allosaurus.wav &

Switch to the "Recognizers" tab and then select "CMU Linguistic Annotation Backend" from the Recognizer dropdown list at the top and then click the "Start" button. If this is your first time using this plugin, you will be prompted to login to the CMULAB backend server and get an access token (you can create an account or simply login with an existing Google account).

More detailed instructions for each specific service (phone transcription, diarization etc.) can be found here

cmulab_elan_extension_screencast_2022-03-25-00.00.25.167-00.02.55.311-seg2.mp4

---

Note: if the "Parameters" section in the "Recognizers" tab is too small, you can pop it out as a separate window by clicking the button below it or by dragging the section border to re-size it. See the highlighted areas in the screenshot below: