FineTracker: Getting Started


The Fine-Tracker (ftracker) software package can be downloaded: version 1.2 and version 1.0.


To unTar

gzip -d ftracker.tar.gz

followed by:

tar xvf ftracker.tar

To run

java -jar ftracker.jar

It will then explain its usage itself. Note that at least Java Runtime Environment 6 is needed.

Test files

The following files are provided to run a test with Fine-Tracker:

  • test.input: This file consists of three utterances, each consisting of a single Dutch word:
    1. kandelaar.input (English: candle)
    2. kijker.input (English: spectator)
    3. cocktail.input (English: cocktail)


  • The name of the file (without .input) is the word spoken in the file. The input is created using multi-layer perceptrons (MLPs) trained on a large corpus of Dutch read speech. More information on the MLPs and the way the input was created can be found in this paper: Modelling fine-phonetic detail in a computational model of word recognition.
  • test.lex: The lexicon consists of 27,591 Dutch words transcribed in terms of articulatory features. See the Document page for more information.
  • default.xml: The config file that should be used for running the test.
  • default.output: The output file that should be created when running the test.

To run the test

java -jar ftracker.jar default.xml

This call will run Fine-Tracker using test.input and test.lex. The output of Fine-Tracker should be identical to the file ‘default.output’.

Back to Introduction