0

I am currently researching for a (fictional) draft and need help. I am a complete layman in music and musical theory.

A single person or a group of people is singing a song to control a machine, which has a vocal interface.

  • The lyrics are not used in this process
  • They may be backed by instruments, but the chanting is the important part

The machine can be "instructed" to do the following (not complete, here to give you an idea):

  • work slower / faster
  • increase / decrease working intensity
  • end the processing of a single piece of work and go to the next.
  • pause work and save energy (note: silence will stop the machine at once if longer than a "normal" pause in a song, so this cannot be encoded with prolonged silence)
  • Fuse two or more simple pieces of work to a more complex one.
  • There should be a possiblity of "doing it wrong" with dangerous consequences (e.g. malfunction, energy discharge)

What I want to do is to somehow "map" vocal input (as a song) to these machine actions. A properly chanted song is then a chain of working instructions which will produce a fixed outcome.

Which properties of chanting (e.g. volume, pitch, speed, melody, rhythym) can be assigned to this outcomes?

Note: I am a complete layman. Should I choose other properties of a song?

How could the structure of a song (recurring themes etc.) be used to encode information?

Are there "laws" for a harmonic song, which if not obeyed could produce suboptimal to very dangerous results?

Is this the right stack exchange to ask this question?

EDIT: Thanks to David Vogel
This is not for a practical application. The machine can do whatever I want it to do. My question is about which properties of a human chant are best suited to transfer "reliable" information.

|improve this question
1

It's a little unclear to me what you are trying to do with this, but here are some thoughts:

Pretty much any audio property can be used to control this system granted you have the ability to analyze the input and know how to map it to the control value.

If you do not want/need the input filtered out, you could use a glockenspiel, for example, which will produce a very identifiable amplitude and frequency change which can then be mapped to the control value you desire.

If you cannot have a noticeable change in the sound like that, it is much more difficult. You would have to use an identifiable change in pitch (frequency), rhythm (time), amplitude (volume), or timbre (frequency spectrum--probably something other than a voice).

|improve this answer
  • This is not for a practical application. The machine can do whatever I want it to do. My question is about which properties of a chant can be used to tranfer information. Thanks for the valuable input! – user31164 Jul 24 '16 at 13:29
  • In that case, the last paragraph gives you the properties of a chant which could be used. – David Vogel Jul 24 '16 at 13:39
  • Would you mind to elaborate on the last part? How would you advise to map this properties to an outcome? Are these 4 properties all there are, or could you point out some more? Are there "laws" of harmony? I know this may be highly subjective, but I find this hard to form out properly. – user31164 Jul 24 '16 at 14:11
  • 1
    The main components of a chant are the melody/harmony (frequency and frequency spectrum), the way it progresses through time (rhythm), and the "loudness/volume" (amplitude). The values for these components can easily be captured and analyzed. It would be up to you how you would want to map these values to control other aspects of the project. There are no "laws" unless you are trying to adhere to the conventions of some specific style. – David Vogel Jul 24 '16 at 15:10
  • 1
    It might be useful to do some historical research on uses of music to communicate information between humans, for example military bugle calls, and various type of signalling horns (e.g. Swiss alphorns) that were used for communication in mountainous regions of Europe. There is no obvious reason why a machine couldn't be built to recognise the same type of messages. – user19146 Jul 24 '16 at 23:22
1

You can pretty much do all the things that are used in digital communications (theory and practice). For examples:
Code by pitch , aka the tune.
Code by rhythm, i.e. half, quarter, eighth, etc. notes act as Lev-Zimpel compressed data bits.
Code by meter - which is closer to FM broadcast, sort of, in that you're sending information over the speed rather than the "carrier" tune.
Code by amplitude, rather like AM broadcast. Or any combination of the above.

Further, dunno if you've heard of the method used in cable/internet traffic: "QAM" , or quadrature amplitude modulation. This basically sends the same data twice, out of phase, for error correction. A "canon" would do this :-)

|improve this answer

Your Answer

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