Reprocess all tracks using updated ffmpeg script

added by @auggod, jan24.2021

[ ] Use backblaze b2 api to download original files
[ ] Do a quality check: validate formats (flac, wav, aiff only), check if metadata is present (related: Trello)
[ ] Conversion with updated ffmpeg script (https://gitlab.resonate.ninja/resonate/upload/blob/develop/api/src/jobs/convert-audio.js)
[ ] AAC (.m4a) files should be stored to separate volumes (10 TB max each on hetzner cloud) accessible to upload tool servers

Current archive is around 1TB. (1,4TB including processed files)
The script is configured to remove any video stream.

The major difference with the new script is that we removed ‘High-Efficiency AAC’ options. Many devices don’t support ‘HE-AAC’. More details here: Encode/AAC – FFmpeg

If later we need a farm of upload tool servers with many data volumes we should look into hetzner networking options to make data sharable between different servers. We’ll need this feature soon to allow new api server to have access to the data for streaming.

Trello link info (was archived): Scan database for MP3s and other files that shouldn’t be there. We should only stream the same, consistent file format in order to debug problems with browsers/devices.