I\'m running Google Cloud Dataflow etl pipeline over a (bounded) collection of audio files hosted in a single GCS bucket.
The preprocessing script inside the pipeline