NU-Aligner

NUaligner is a GUI (Graphical User Interface) wrapper built around the SONIC toolkit developed by CLEAR - Center for Computational Language and EducAtion Research (formerly CSLR - Center for Spoken Language Research)

NUaligner and the SONIC toolkit provide a platform for quick automatic word/phoneme alignment of speech files that have already been sentence aligned. NUaligner takes the sentence aligned speech and converts produces a phoneme/word aligned praat text grids. The text grids can be inspected in Praat and hand corrected accordingly.

Using NU-Aligner

  1. Create a folder containing the transcription text files and .wav files. All files must have the same prefix name. For example, you folder may contain the following:
    • speechfile001.wav
    • speechfile001.txt
    • speechfile002.wav
    • speechfile002.txt

    The format of each transcription text file is as follows:
    START_TIME [TAB] END_TIME [TAB] TRANSCRIPTION
    An example file can be downloaded here

    Wavescroller can be used to transcribe spotaneous or scripted speech.

  2. Launch NUaligner.
  3. Click "Select Sound File" to process a single file or "Select folder" to batch process a whole folder. If you click "Select Sound File", navigate to the .wav or .aiff file that you wish to process (speechfile001.wav or speechfile002.wav).
  4. Check the "Phonetic Alignment" checkbox if you wish to obtain phonetic alignment of the speech files. Leave unchecked if you wish to have word alignment only.
  5. Click "Go".
  6. Once the files have been processed you will see a .TextGrid file in the current folder (e.g. speechfile001.TextGrid).
  7. The .Textgrid file can be opened in Praat and will contain the word/phoneme aligned boundaries of the related speech file.