Are there conventions for describing the visual representation of sounds on a spectrogram (i.e. names for shape patterns)?

Question

I have a growing set of bioacoustic recordings with accompanying spectrograms from marine animals in an online repository. Currently, the recordings can be searched based on the sound 'name' as applied by the researchers who captured the recording, usually as reported in an academic reference. These terms are descriptive of the sound in an auditory sense, but assessing what a sound might reasonably be called is subjective both when it is first named, and when a user trying to identify a mystery sound has to guess what someone else might have labeled it.

In order to make the recordings more searchable, I would like to create labels that describe the spectrograms visually, as in based on a general pattern of shapes the plot produces. Below are four examples of spectrograms that illustrate the issue. In terms of the sound as heard by the ear, and the sound label given to each, they are very different. But in terms of the visual representation, they could all reasonably be described as 'column of stacked horizontal lines'.

For a sighted person trying to identify a mystery noise, I feel that this style of visual description would be much easier to use for searching than an acoustic label. (Note that the acoustic labels would remain, and that I also hope to add searching by other acoustic features such as frequency when I can determine a method to reliably extract this information from each file). I am uncertain, however, whether there are already conventions for naming spectrogram forms this way.

Are there any known standards, conventions, controlled vocabularies, notable examples etc. for describing the visual patterns of a spectrogram for a bioacoustic recording, and in which disciplines or areas are they used?

Answer 1

Might I suggest the site earbirding.com/blog/specs or the book Peterson Field Guide To Bird Sounds Of Western North America (Peterson Field Guides). While these are used for birdsongs in the given examples, they may help in describing marine sounds in spectrograms. Hope this helps!

[Please note that this was first posted as a comment.]

Answer 2

I would look a bit in the bird world for this. There are many visual based ways of describing sounds. Bill Evans' Old Bird is a great example, as his classification of night flight calls includes upsweep, downsweep, double up, buzz, etc. I know the same is true for birdsong, but I am less familiar with where to look.

Edited to add, I think this is a great idea and I hope you are successful!

Answer 3

I consider it challenging to obtain unique or commonly accepted description of spectrogram features.

For example very fast clicks appear different as function of analysis window. They can in the extreme look like a tonal with or without harmonics. An observation that was already made by WA Watkins harmonic interval paper

I agree with @Shannon, to recreate an on-line repository that allow discussion on how to best call these features

Ref:

Watkins, William A., The harmonic interval: fact or artifact in spectral analysis of pulse trains. WHOI Technical report, 68-13, 1968.

Answer 4

For the bird world, Nathan Pieplow's book mentioned above really set new standards in terms of communication about vocalisations. I'd recommend to check it out too.

Answer 5

Bioacoustics Stack Exchange!

I might suggest that the Bioacoustics Stack Exchange might be a good place to get a wide variety of community input on appropriate descriptive terms.

I would recommend that this be done separately for each type of call (and in an ideal world this would include similar sounds from different taxa), but this variety could also be represented in the answers people present.

This question could also be phrased such that you seek recommendations for both descriptive term(s) for a given call, but also for more formal/scientific terminology (as this is far from consistent!).

Your question here is looking for an online library (so I would recommend separating it from the implied question of what descriptions are appropriate for this particular type of call). I suspect there may be different common descriptors for different taxa.

If ultimately you agree (and the community agrees) with this answer-- then I would recommend editing your question to ask about an online repository, and then re-ask the separate question regarding that particular call type. I would also recommend asking for both common descriptive term(s) as well as more formal/scientific terminology (as this is not as consistent as one might hope!).