Acoustic Recording File Description and Annotation Standards - What Exists? What Should Exist?

Question

I am hoping to compile a list of existing standards or recommended approaches for describing acoustic files, including any notable strengths or weaknesses with the systems.

My team has access to large inventories of acoustic soundscape recordings (hopefully I am using that term correctly; unprocessed multi-hour recordings) that have been described in different ways depending on when, how and by whom they were captured. We are hoping to standardize the metadata describing them to make it easier to locate and use files of interest. If appropriate, we can establish our own recommended system, but 1) don't want to further complicate the space if suitable standards already exist, and 2) would like to ensure anything we did make addresses known issues.

Does anyone know of standards or approaches worth considering? I suspect there may be three 'levels' or types of systems out there:

• How to describe soundscapes or the original full-length recording
• How to describe recording clips extracted from x-y time points in a full length recording
• How to describe an annotation for a known or unknown noise in a clip

We are most immediately interested in soundscape examples, but having a convenient list of each might be useful for the community.

Answer 1

The short answer there is not one standard and more work is needed.

For encoding metadata in an audio file, see Guano:https://github.com/riggsd/guano-spec/issues

For describing audio files and their contents see: https://www.tdwg.org/standards/ac/

Personally I think Audobon Core is too heavy for the common workflow.

Open Ecoacoustics has a incomplete draft on some more informal standards here: https://github.com/ecoacoustics/metadata-standard

Answer 2

The only thing I know is the Universal Category System. However, it is intended for the classification of sound effects and therefore may not be perfectly suited for bioacoustic needs.

It is essentially just a framework for the consistent categorization of sounds using a defined set of categories, subcategories, etc.. Its standardized form of meta information is intended to quickly find sounds in huge sound databases. As such, it may only serve as a starting point for developing a framework that is more appropriate for bioacoustics.

Answer 3

Tethys is worth mentioning https://tethys.sdsu.edu/

Quoting their website:

Tethys is a freely available open source temporal-spatial database for metadata related to acoustic recordings. The database is intended to house the metadata from marine mammal detection and localization studies, allowing the user to perform meta analyses or to aggregate data from many experimental efforts based on a common attribute.

However, Tethys is heavy and not necessarily flexible enough for all applications, and therefore I believe the discussion of a new metadata standard is warranted.