recognition of intonational phrase boundaries

[Monika Dominguez]

Hello,

We are starting to use AuToBI following the manual's directions but we do not get the expected results. We are using a short sound file and textgrid with the word by word segmentation. Then we are using the manual's command line:

java -jar AuToBI.jar \
-input_file=<filename> \
-wav_file=<wav file> \
-normalization_parameters=<parameters filename> \
-pitch_accent_detector=<model_filename> \
-pitch_accent_classifier=<model_filename> \
-IP_detector=<intonational phrase detector filename> \
-ip_detector=<intermediate phrase detector filename> \
-phrase_accent_classifier=<model_filename> \
-boundary_tone_classifier=<model_filename> \
-out_file=<filename>

But the output only shows accents word by word. It does not detect intonational phrases, and thus pitch accents and boundary tones. (See attached file with the output)

How can we get the right format taking into account intonational phrases and not words as labelling units?

Thank you in advance for your help

Regards,

Monika