With the aim of training machine learning algorithms for vocal mode classification, I am collecting audio recordings of singing voices in any of the four vocal modes defined by the complete vocal technique (CVT), i.e. neutral, curbing, overdrive and edge. Anyone familiar with CVT* is welcome to supply audio recordings of themselves singing in one of the vocal modes. To participate, you could
- Record scales (major, chromatic,...) singing the CVT mode vowels (or, if you dare, any vowel of your liking)
An example (of myself) showing scales in neutral and overdrive (mostly) can be found here and here. The easiest would be a simple chromatic scale covering some of your range.
A duration of at least 1 second per note would be awesome. It would be great, if the recordings did not contain other instruments or considerable background noise. Do not worry if there is a note or two out of the target mode, I can just cut them out (or relable them) later. You do not have to be pitch perfect, correct pitch is not required. The entire range of the human voice is of interest, be it high or low notes, breathy (audible air), brighter/darker, with/without vibrato, reduced/full density, ... All that matters is the vocal mode. Be as creative as you like.
Because I would like to actually make use of the trained algorithms, and to publish their results, I require your consent to publishing results that make use of your data. Separately, optionally, your recording could become part of a dataset I am setting up (and would eventually publish). If you would find this acceptable, check "Yes" at the last question. Everything you supply is fully anonymous. I do not record IPs or anything, all I see are the responses you give plus the uploaded audio.
If you have any question, feel free to contact me under CVTVocalModeResearch@tnt.uni-hannover.de
Thank you very much for participating :)
* Unfamilar singers of sufficient profiency could attempt to replicate the CVT vocal modes, examples can be found online, e.g., here.
