Android上的声波通信

Question

I want to transfer data by sound wave on android phone, and I find this example : https://github.com/JesseGu/SinVoice

this project work fine, when it transfer data, I can hear sound and transfer data success.

Problem is how to modify code for it transfer data but i hear no sound ?
If is used high frequency sound, please tell how to modify it ,thanks.

Answer 1

Chirp for iPhone is a cheap way out of solving a difficult problem. They actually upload the data to be sent online, the "chirp" appears to just be a unique sequence of notes which indexes to the uploaded data on the online database. That being said, I have been interested in genuine, audio data transmission for quite some time now. If you're going to do this project, you need to worry about the code yourself, what I can show you is a bunch of technologies you may want to incorporate into your design. My advice would be to start from scratch, but you could try building off of this Github code if you really want to.

For detecting and filtering out sounds, use the Goertzel algorithm. For an input sound, the Goertzel algorithm filters the stream of incoming sound samples and produces a complex valued stream of output values ( a+bi ). The magnitude ( sqrt((a^2)+(b^2)) ) of the values gives the intensity of a particular frequency at the given point in time. The angle ( atan2(b,a) ) of the values gives the phase of that frequency at the given point in time. You can choose which frequency you want to filter on and, in particular, use multiple instances of the Goertzel algorithm to detect multiple frequencies.

http://www.mstarlabs.com/dsp/goertzel/goertzel.html

You can simply play a pattern using one tone, being on or off, filter the sound coming in through the microphone using the Goertzel algorithm, and use that to detect the pattern played, thus recovering the bits of information you sent. This technique is a simple form of amplitude modulation and is very susceptible to background noise. Your data transfer rate will be very slow.

https://www.princeton.edu/~achaney/tmve/wiki100k/docs/Amplitude_modulation.html

You can filter for several different notes and have your program listen for sequences of notes. This is most likely how Chirp works internally, and this is very reliable. Your data transfer rate can be somewhat improved.

You can also use frequency modulation or phase modulation. Using the Goertzel Algorithm will mean phase modulation is somewhat easier than frequency modulation. I have not tested this approach as extensively, but it appears to be incredibly reliable and the bit rates could be higher than amplitude modulation. I once wrote a program that sent pictures using phase modulation. The data was not digitized, I just determined pixel intensity by phase and scanned across all of the pixels in my image. The picture came out fairly clear even when playing loud music. Note that the images were fairly small and it still took a couple minutes to send, the bitrates for all audio communication are going to be pretty low.

http://www.circuitstoday.com/modulation

One quick word of advice. Consider these two ways of calculating a sine wave to play

double output[length];
double t=0;
double samplerate=44100.0;
double frequency=440.0;
double w=2*pi*frequency;
double time_increment=1.0/samplerate;
for(n=0;n<length;n++) {
    output[n]=sin(w*t);
    t+=time_increment;
}

Second way

double output[length];
double samplerate=44100.0;
double frequency=440.0;
double phase=0;
double phase_increment=2*pi*frequency/samplerate;
double tau=2*pi;
for(n=0;n<length;n++) {
    output[n]=sin(phase);
    phase=(phase+phase_increment)%tau;
}

For the first way, when you have large values of t, the value inside the sin will lose precision. Essentially, the longer you play the note for, the crappier it sounds. After not too long, it begins to sound very impure.

The second way avoids this problem by integrating the phase over time and modding (%) it by 2*pi (tau), keeping the value in the range 0 to 2pi.

Answer 2

Here's an Android implementation of Chirp , which maps data to frequencies of sound. In addition, to mitigate data loss, transmitted frames of data are supported by error correction symbols.