The following document allows you to Slurm documents. There are two parts.
This includes submit.sh, template.sh, QBASR_GenerateData.py
- Add file in proper json format from https://sites.google.com/view/qanta/datasets?authuser=0
- Update QBASR_GenerateData to save in the right place
- Update template.sh to refer to the proper anaconda path, file, etc.
- Run "submit.sh 48" (or appropriate amount of partitions)
This includes submit_decode.sh, template_decode.sh, QBASR_GenerateFileList, QBASR_DecodeAudio.py Depends on kaldi installation and ffmpeg (to make .wav files from .mp3s)
- Run QBASR_GenerateFileList on the proper folder e.g. "/dev/*.mp3"
- Update template.sh to refer to proper file and DecodeAudio to write to correct location
- Run submit_decode.sh