Prosograph2 is an extension on Prosograph for viewing bilingual speech corpora. Aligned samples are displayed side by side to accommodate e.g. prosodic comparison. The tool also reads prosodically annotated speech data from Proscript format files.
Prosograph is written in Python mode of Processing. Once you downloaded and installed Processing, put it in Python mode and then open the main script Prosograph.pyde
to run.
minim
library needs to be installed for playback. You can do this directly from Processing.
Prosograph reads Proscript files. Sample dataset is provided in data
directory.
Data specific configurations are set in dataconfig_<dataset>.py
. Inside this script you need to define DATASET
as the path to your proscript file or the directory containing your proscript files. You also need to specify the keys used in your proscript files. The configuration file needs to be imported in main .pyde
file on line:
import dataconfig_heroes as dataconfig #Specify here which dataconfig you want to use
To view sample datasets, make sure you change DATASET
to the exact directory where the dataset resides.
Visual configurations are set in config.py
.
Prosograph is operated with the following hotkeys:
N
- Skip ahead samplesB
- Skip back samplesP
- Play/pauseR
- Refresh viewS
- Save current view as image to diskC
- Change colour paletteX
,Q
- Exit
For playback, audio file needs to be in the same directory and with same name as the proscript file. To play, a section needs to be selected first using left-clicking on wordboxes.
Sample data in this repository is taken from Heroes corpus
To create your own proscript files, please refer to Proscript python library.
If you use this software, please give attribution. If you want to cite it in your work, you can use the following bibliography entries.
@article{lre:prosograph,
author = {Alp Oktem and Mireia Farrus and Antonio Bonafonte},
title = {Corpora Compilation for Prosody-informed Speech Processing},
journal = {Language Resources & Evaluation},
status= {accepted}
}
@inproceedings{interspeech2017:prosograph,
author = {Alp Oktem and Mireia Farrus and Leo Wanner},
title = {Prosograph: a tool for prosody visualisation of large speech corpora},
booktitle = {Proceedings of the 18th Annual Conference of the International Speech Communication Association (INTERSPEECH)},
year = {2017},
address = {Stockholm, Sweden}
}