When I hear a sound, I have three ways to describe it:

  1. If it has a real source I just call it. (e.g. Piano sound)

  2. If I know the method to produce it(sawtooth osc to lp filter with cutoff linked to a LFO)

  3. If I could mimic it (like wahwah)

However these methods are usually not convertible. Say if I have a sound that I can mimic, then usually I don't know how to create this sound.

This brings difficulties and confusions when I want to create an index for all the sounds I know (and to make/discover new sounds).

So is there a better way to do this?

Also, it would be wonderful if there is already such an index. :) Can someone give me some hints?



interesting question! Yes, there are systemized/categorized indexes of sounds, these are called a "thesaurus" in libraries and archival practice. It's described as "a book that lists words in groups of synonyms and related concepts." Of course you don't have just 1 book, and many have different rules.

What I think is your issue is not a lack of words/ways of describing sounds, but the very nature of sound. Sound is a very subjective phenomenon. With that I mean that a lot of people hear different things, when presented with 1 sound. This is a good thing btw, because it allows people to have similar experiences instead of identical ones.

What you could do is think about what emotion a sound tries to evoke and perhaps add that to your index. Or what use you think the sound will have.

However I strongly believe that there will never be 1 or 2 fixed indexes. Culture changes and hence so does the 'meaning' of a sound.

  • You have shed precious light into my dark confusions. Thank you! Yes, there are many 'books', some about audio techniques, some about genres, all have different rules. It's like I have a file and many folders and I don't know where to put it. You mentioned the nature of sound. When people have similar experiences with a sound, it becomes a genre(of course as you said, with some variations), in my opinion. A genre sound index might be a good start, but it's still coarse-grained. It's like the libraries you have in synths, having 'pads', 'solo leads' and so on, but with no emotional tags. – Yadli May 14 '14 at 15:09
  • Going back to my 'files' metaphor, a file might belong to multiple folders. I can create 'duplicates' right? :-) Also, I guess it's like neurone networks in computer science where an implicit layer exists to control some internal states that will affect the final representation. That could be 'emotion triggers' in musical sense. I even wonder if this representation can be combined with engineering techniques -- extract features (as @Sinister pointed out), link them to implicit stuff, and when they are linked together, a musical concept will come out, like 'deep' sometimes means 'reverb' right? – Yadli May 14 '14 at 15:15
  • Your files metaphor kind of sounds like a tagging system? A sound can have many tags, and a tag can belong to many sounds (in relational database parlance). – JoshP May 14 '14 at 15:29
  • @JoshP yes, something like that. But what will happen if tags can be tagged? Then this becomes a two-layer system, unlike a flat relational table. Taking it further, if we could make links between anything(thus creating paths with infinite length), we could build a graph database. It's like a social network of sound concepts! :-D This would be powerful... FB has a feature to "find the friends of your friends". What if we could do this with sounds? First we can expand the related concepts. Then if we go further(into the second hop) we can even make variations to create something completely new. – Yadli May 14 '14 at 16:20
  • It sounds like you want a tagging system with 'families' in it, that is already available in most software apps. Genre is one example. With that you can 'tag' things as 'distorted' and also add a family (or meta-tag) 'crunched' or 'scary'. However I would try to keep things a bit separated and create/think up tags and the family names from two different perspective. For example, tags are descriptions of the sound-colour or timbre, whilst family names are named after possible usage based (example: "horror,sci-fi")... so that already exists.. :) – Arnoud Traa May 14 '14 at 18:50

You could try using something like Sonic Visualizer / Annotator to see and extract features of the sound http://isophonics.net/SonicVisualiser

  • It looks very nice, thank you! I will definitely try it out. It seems the analysis toolbox of this program is very rich(yes Queen Mary!). – Yadli May 14 '14 at 14:57
  • However, this only solves part of my problem(see option 2 in my post, the engineering way, which I love) -- and with this in mind, I cannot communicate with others sometimes. For example, the sound created in option 2 would be referred as 'dubby bass', but I might not have acquired this word, especially if it's not 'dubby' but rather some less popular genres. – Yadli May 14 '14 at 15:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.