build pre-filled eaf based on previous diarization annotations

LAAC-LSCP / ChildProject

Python package for the management of day-long recordings of children.

https://childproject.readthedocs.io

MIT License

13 stars 5 forks source link

build pre-filled eaf based on previous diarization annotations #334

Closed lucasgautheron closed 2 years ago

lucasgautheron commented 2 years ago

[x] tests
[x] documentation
[x] implementation

lucasgautheron commented 2 years ago

Example:

child-project eaf-builder --destination ../test/eaf --segments ../test/segments_20220110_174937.csv --eaf-type random --template basic --path . --import-speech-from textgrid/mm

lucasgautheron commented 2 years ago

FYI, the only thing that prevents merging at this step is the lack of test coverage for this feature! Will do it ASAP.

LoannPeurey commented 2 years ago

additional modifications :

prefix imported tiers with the name of the set (eg set vtc for speaker CHI => tier named VTC-CHI
replace NA values for speaker_type by SPEECH for some specific formats (vtc_rttm and vcm_rttm)