Blizzard Challenge 2016-8 Git Repository
Some, but not all, of the material has been aligned with the book text. Because some of the books provided by the publisher are in PDF format, and the audio for each book is typically a sequence of audio tracks, it is non-trivial to extract a clean version of the text and then to align it with the speech (e.g., at sentence level).
Therefore, we ask all participants to collaborate and share the effort of creating clean, sentence-aligned transcripts of the speech. This will be co-ordinated by Sebastien Le Maguer. Please contact Sebastien before embarking on any manual cleanup of the text, or alignment with the audio, so that he can eliminate duplicate work.
The shared repository for cleaned-up text and alignments with the audio can be found at http://blizzard.coli.uni-saarland.de/root/blizzard_training_data/blob/master/README.md - please use the same username and password as for the data download.
Please note that any alignments of the text that you commit to the repository should be with respect to the publisher's original complete audio files, as distributed for the Blizzard Challenges 2016, 2017 and 2018.
The text in this repository is copyrighted and only available after signing the data license.
Current contents of the repository
Some of the teams that participated in the 2016 and 2017 challenges have contributed to the repository, as detailed below. Please can all other participants please consider also contributing!
We expect teams to continue contributing to this repository during the 2018 challenge. This includes teams from industry!!
- Branch: CSTR
- http://blizzard.coli.uni-saarland.de/root/blizzard_training_data/tree/CSTR
- init repo + cleaning text
- Branch: NII
- http://blizzard.coli.uni-saarland.de/root/blizzard_training_data/tree/NII
- manually corrected text
- Branch: INNOETICS
- http://blizzard.coli.uni-saarland.de/root/blizzard_training_data/tree/INNOETICS
- sentence-level alignment which includes:
- A more consistent use of quotation in dialog parts.
- Square brackets to denote non-linguistic vocalizations. The text within [] is a very rough representation of the vocalization but is not used consistently and should not be considered appropriate for any further processing. It is rather a placeholder that could help the segmentation process work its way around these vocalizations. At least, it works for us.
- Branch: NITECH
- http://blizzard.coli.uni-saarland.de/root/blizzard_training_data/tree/NITECH
- New architecture:
- the text split by page
- the wavs split by page. This part is not versioned but can be retrieved from http://data.cstr.ed.ac.uk/blizzard2017-18/usborne/contrib/nitech_wavbypage.tar.bz2 (requires a license for the 2017-18 data, then use the same username/password)