Updated and Cleaned corpus

anrizal commented 5 years ago

This updated corpus is used as part of final project in Speech Technology Course in KTH (Royal Institute of Technology Sweden).

Our group members are: Arra’di Nur Rizal - Jonathan Jan - Luise Dürlich - Shifei Chen

Updates to the original

All of the data has been converted to CAF or 3GP format with uniform bit and a sampling rate of 128 kbps and 8kHz
Using DBL (Dunstan Baby Language) categories as a reference, data tagged with lonely, scared, and unknown are removed. Additionally, data tagged with cold/hot was merged into discomfort.
All non-cries data has been removed manually by listening to them. These include white noise, baby chat, adult mimicking baby cries. etc.
All data are put into its respected folder without changing the original filename.

gveres commented 5 years ago

@anrizal , this is great, thanks for the PR! Is there a public domain publication with your results?

anrizal commented 5 years ago

@anrizal , this is great, thanks for the PR! Is there a public domain publication with your results?

Hello, the publication link is not out yet. But, the university is planning to open the publication. I am not sure, whether it will be on the university website or in arxiv. I'll create a pull request with link to the publication in the README as soon as its available.

gveres / donateacry-corpus

Updated and Cleaned corpus #3