Instead of noise cancelling do noise adding. Assuming you can predict and calculate the waveform to add to the ambient noise in realtime (low power GPUs, Machine Learning) just add the waveform at the ear in the headphones. Ideally if you used it in a gym or running you could harvest energy to power it from motion, ever present radio waves, thermocouple, etc. and the headphones would have a small rechargeable battery but use no external energy. What are the challenges? What are the potentials?
The idea of adding artificial noise to the ambient noise to produce music is very interesting. Just to clarify, are you proposing a music player that "reacts" to the environment to create new songs on the fly? If it is, there is a player that does something similar, although not by adding noise (although in some variations, it kind of does), but by molding the noise. You should check out the RjDj program and "reactive music": http://en.wikipedia.org/wiki/RjDj
In any case, there's an important challenge here which is that of creation of "structured" music (chorus, verse, etc.) from the environment. RjDj, as well as some easily implementable ideas that come to mind (which don't even require Machine Learning), present music that is "atmospheric" and never-ending, which is understandable, as those are virtues of a typical acoustic environment. However, a good challenge would be to use the surroundings to create structured, finite pieces of music, on-the-fly, that aren't based on a-priori created music.
When it comes to powering them, however, I think that over-the-counter smartphones have enough processing power to carry out any of these algorithms (a good example is, again, RjDj which is an iOS app). Still, your idea of powering them, although separate from the music creation point, I think has legs. Powering from motion, radio waves, and/or thermocouple would be a good compliment to today's handhelds, and be very welcome to those of us that need to carry a charger/2nd battery around.
Cheers,
Caleb
Caleb,
Thanks for the quick response and RjDj tip. I am actually wondering if you can play commercial songs like a regular music player. One downside I thought of is that when people are calling your name to get your attention the device would just use that to add on top of. But there could be a volume limiter that would allow alarms and voices to be heard.
A better refinement of the question would be how much energy could be saved if you were using the ambient noise in a coffee shop or gym or city. Instead of using energy for cancelling out the noise and then playing the song on top of that, add those two waveforms from the beginning.
Thanks,
Brendan
Joachim,
Thanks, this is interesting. Yes this is similar in the sense that you use the existing ambient noise to add on top of.
Now that I understand how the two points come together (add on top of existing ambient noise to reduce voltage consumption while playback), I'm not too sure if the amount of saved energy would be enough to counter the amount of energy required to carry out the intelligent addition of both signals. This is because it would also require a microphone to obtain the environmental noise as well as its corresponding ADC's, and to power some type of FPGA to carry out a predictor and signal smoother to account for acoustic changes in the environment.
In addition, considering that standard line levels for consumer are less than +- 1.5 V (http://en.wikipedia.org/wiki/Line_level#Nominal_levels), I suspect that the majority of energy consumption is NOT in the physical delivery of the audio signal (DAC), but in the sampling of files and/or in powering its memory.
Still, the problem of predicting/smoothing is interesting because of the challenge.
Cheers,
Caleb
- Badges
- Science topic