Everything to built an ASR for Wolof (standard)
## Wolof Data collected by Elodie Gauthier & Pascal Uriel Elingui
### Prepared by Elodie Gauthier & Laurent Besacier
#### GETALP LIG, Grenoble, France & Voxygen SAS, Dakar, Sénégal
#### OVERVIEW
The package contains Wolof speech corpus with audio data in the directory *data/*. The data directory contains 6 subdirectories:
1. **train** - speech data and transcription for training automatic speech recognition system (Kaldi ASR format 1 )
2. **dev** - speech data and transcription (verified) to evaluate the ASR system (Kaldi ASR format)
3. **test** - speech data and transcription (verified) for testing the ASR system (Kaldi ASR format)
4. **dev_unverified** - original speech data and transcription (NOT verified, contains mispronunciations)
5. **test_unverified** - original speech data and transcription (NOT verified, contains mispronunciations)
6. **local** - for now, contains the Wolof vocabulary (one without vowel length contrast, and another one containing vowel length contrasted units). Once you will ran the *run.sh* script it will contain the *dict/* and *lang/* directories needed to build the ASR system.
*LM/* directory contains 2 text corpus, the language model and its perplexity computed from the *dev* and *test* datasets.
#### PUBLICATION ON WOLOF SPEECH & LM DATA
More details on the corpus and how it was collected can be found on the following publication (please cite this bibtex if you use this data).
@inproceedings{gauthier2016collecting,
title={Collecting resources in sub-saharan african languages for automatic speech recognition: a case study of wolof},
author={Gauthier, Elodie and Besacier, Laurent and Voisin, Sylvie and Melese, Michael and Elingui, Uriel Pascal},
year={2016},
organization={LREC}}
### SCRIPTS
In *kaldi-scripts/* you will find:
* **00_init_paths.sh** - it initializes your PATH variable (required to run the Kaldi scripts)
* **01_init_symlink.sh** - it creates the symbolic links (required to run the Kaldi scripts)
* **02_lexicon.sh* …