Data structure
To detect chirps on a recording of single or multiple electrodes, the dataset must fulfill the following requirements:
- The dataset is a single directory containing subdirectories for each single
.raw
or.wav
file, i.e. for each recording. - Each raw recording file, i.e. the
.raw
or.wav
file is namedtraces_grid1.{file_ext}
. - The fundamental frequencies of the fish in the recordings must be already tracked using the wavetracker project.
An example directory structure could look like the following:
dataset_root
├── chirpdetector.toml # the chirpdetector config
├── 2019-11-25-09_59 # dir of first recording
│ ├── fund_v.npy # wavetracker
│ ├── ident_v.npy # wavetracker
│ ├── idx_v.npy # wavetracker
│ ├── sign_v.npy # wavetracker
│ ├── times.npy # wavetracker
│ └── traces_grid1.wav # raw recording
├── 2019-11-26-09_54 # dir of second recording
│ ├── fund_v.npy # ...
...
If these requirements are met, detecting chirps should be no issue.