Protege, post: 377028 wrote: If a very short sound was to be produced by an ordinary speaker, would humans hear this? Could this be done so that the sound playing at regular intervals is detectable by a device but is not irritating for humans and pets? Does this depend on frequency and/or intensity and how?

If it's produced by an "ordinary speaker", it's likely to be audible to most people, since most "ordinary speakers" are designed to reproduce frequencies within humans' hearing range. But, it does depend on how loud it is, how far away the ear is, and the person's hearing range and sensitivity. You may have some speakers that can reproduce higher frequencies than most people can hear, but I wouldn't label them as "ordinary". Some people can hear slightly higher than 20kHz, and if it's higher than that, at a low volume, and a very, very short instance, most people may not hear it, and even some people with exceptional hearing may only...what's the word...'perceive' it?

"Very short" also leaves the possibilities of a zillion variables to go with all the other variables of level, frequency, etc. How long is "very short"?

Humans and animals have different hearing ranges. That's why there are dog whistles. If you were to make something that absolutely no pets could hear, it's doubtful that any human could hear it.

What's this about, anyway?

Kapt.Krunch

"Very short" sounds have onset and cessation times that themselves produce audible clicks, independent of what the actual short sound is meant to be. However, it could be that if you were to encapsulate a single cycle of a high-frequency sinewave, starting and finishing at zero, this would not be heard, but it would also not be easy to detect by a device for signalling operations.

A better scheme would be to look at the sort of encoding done in FM stereo radio, where a low-amplitude (10% modulation) 19 KHz pilot tone is continuously present during stereo transmissions. This 19KHz tone is filtered out of the audio, and is used to lock the phase reference of the receiver's stereo decoder. The presence or absence of the pilot tone switches the decoder on and off, and there is no reason why this switching could not be used to control devices other than a decoder. Whether the transmission of a 19KHz sound through a loudspeaker to a microphone would be reliable enough to give consistent results using this technique is a different matter.

You would have to give us more details of what you want to use this signalling for in order for us to give more concrete suggestions.

Objective

Thanks very much for the replies guys. The question is in regards to understanding digital watermarking and I thought this might be a simple way to detect content through an audio watermark. My intention is to be able to do this through a mobile phone as a detection device. As I researched this a bit more I came acrosscompanies like Digimarc or Civolution who mention they have hundreds of patents on this. Now, I am trying to make sense of what might be patented or too complex here.

Hmm. Standard mobile phones have an audio bandwidth of 300Hz to 3.4KHz and a restricted dynamic range. I would be surprised if you could get an inaudible watermark to be recognised through that medium.

Every kind of transmission method has a certain bandwidth. While we commonly know that human hearing goes from 20 Hz-20,000 Hz that's not really true. While most of us can't hear the difference between 15,000 Hz & higher, there isn't any real fundamental musical information there that is 100% necessary to hear. For instance, FM radio has a frequency response of only 15,000 cycles. As Boswell indicated, 19 kHz along with 38 kHz is continually present. Now most of the equipment that we use is all capable of extending out to 20 kHz and beyond. So let's say, I could record myself saying " this is a RemyRAD recording". I would loop that and speed it up in pitch to around 19 kHz. Then you could lay that in over top of your stereo mix at a reduced level. It would probably never be heard by anybody as nothing more than a 19 kHz like tone. But when slowed down substantially, it might create an actual audible watermark to prove your ownership? I've never tried that but when you realize that there are plenty of people out there listening to music that has plenty of 15 kHz TV flyback transformer noise in the audio signal and nobody notices it, what's that tell you? Look at a lot of commercial recordings on a software spectrum analysis. At all of the quiet passages and pauses between movements, you see that it is above the noise floor but generally lower than audibility i.e. -60 DB. But it's there and it's there all the time. And for most people over 40, nobody can hear much beyond 15 kHz. You could also try the inverse with extremely low frequency content but I feel that would be more noticeable. I mean there are a lot of rock 'n roll hits and others out of NYC that all have subway sub harmonics on the track and nobody has ever said anything about it since they don't generally perceive it.

And it comes out there do do do do...
Mx. Remy Ann David