This is a simple flask-based web-application meant to be run locally to quickly visualize/listen to some audio files & view predictions from your model to get an idea of it's performance & error pattern.
- Install python prerequisites
- Run
python prediction_explorer.py PREDICTIONS.tsv WAV_DIR
PREDICTIONS.tsv
has columns(start_time_s, duration_s, confidence, wav_filename)
WAV_DIR
is the directory containing the wavfiles in the TSV
- Follow instruction on the page, and see comments in py file for more info.
To run an example, you could simply use one of the Pod.Cast test sets. (these are already in the above format)
You can download it using AWS CLI with:
aws --no-sign-request s3 cp s3://acoustic-sandbox/labeled-data/detection/test/OrcasoundLab09272017_Test.tar.gz [DESTINATION]
Extract the archive and point the script to the files as in step 2. above.