A simple python script to analyze microphone data, and some javascript for detecting note patterns. The note patterns will be intercepted as node-ar-drone commands, and passed to the nodecopter.
Requires: nodejs (+ ar-drone) python (+ pyaudio, numpy)
Running the application:
python noteanalyzer.py | node ukulelecopter.js
| #!/usr/bin/env python | |
| import pyaudio | |
| import struct | |
| import numpy as np | |
| import math | |
| import sys | |
| FORMAT = pyaudio.paInt16 | |
| CHANNELS = 2 | |
| RATE = 44100 | |
| BLOCK_DURATION = 0.05 | |
| FRAMES_PER_BLOCK = int(RATE*BLOCK_DURATION) | |
| NORMALIZATION = (1.0/32768.0) | |
| MIN_RMS = 0.020 | |
| NOTE_PPRECISION = 10 | |
| #NOTES = [[440, 'A'], [329.6, 'E'], [261.6, 'C'], [392, 'G']] # ukulele, G tuning | |
| NOTES = [[440, 'A'], [329.6, 'E'], [310.0, 'D'], [392, 'G']] # ukulele, vegard-tuning | |
| pa = pyaudio.PyAudio() | |
| stream = pa.open(format = FORMAT, | |
| channels = CHANNELS, | |
| rate = RATE, | |
| input = True, | |
| frames_per_buffer = FRAMES_PER_BLOCK) | |
| def get_rms( block ): | |
| ''' | |
| http://en.wikipedia.org/wiki/Root_mean_square | |
| ''' | |
| sample_size = len(block)/CHANNELS | |
| samples = struct.unpack("%dh"%(sample_size), block ) | |
| sum_squares = sum([math.pow(sample * NORMALIZATION, 2) for sample in samples]) | |
| return math.sqrt( sum_squares / sample_size ) | |
| def frequency(block): | |
| ''' | |
| Resolve frequency by fourier (fft) and quadratic interpolation on data sample | |
| ''' | |
| data = np.array(struct.unpack("%dh" % (len(block)/CHANNELS), block)) | |
| data = abs(np.fft.rfft(data))**2 | |
| fftmax = data[1:].argmax() + 1 | |
| # quadratic interpolation | |
| if fftmax != len(data)-1: | |
| y0, y1, y2 = np.log(data[fftmax-1:fftmax+2:]) | |
| x1 = (y2-y0) * 0.5 / (2*y1 - y2-y0) | |
| freq = (fftmax+x1) * RATE / FRAMES_PER_BLOCK | |
| else: | |
| freq = fftmax * RATE / FRAMES_PER_BLOCK | |
| return freq | |
| while True: | |
| try: | |
| block = stream.read(FRAMES_PER_BLOCK) | |
| rms = get_rms(block) | |
| if (rms > MIN_RMS): | |
| freq = frequency(block) | |
| for hz, note in NOTES: | |
| if abs(hz-freq) < NOTE_PPRECISION: | |
| sys.stdout.write(note) | |
| sys.stdout.flush() # so other proccess can read stdout after each note | |
| except IOError, e: | |
| print("(!) Error recording: %s" % (e)) |
| #!/usr/bin/env node | |
| //var client = require('./dronemock.js'); | |
| var arDrone = require('ar-drone'); | |
| var client = arDrone.createClient(); | |
| var INTERVAL = 800; // ms | |
| var seq = ""; | |
| process.stdin.resume(); | |
| process.stdin.setEncoding('UTF-8'); | |
| process.stdin.pipe(process.stdout); | |
| var patterns = { | |
| 'A' : { 'cmd' : function() { client.land() } }, | |
| 'E' : { 'cmd' : function() { client.left(0.5) } }, | |
| 'D' : { 'cmd' : function() { client.right(0.5) } }, | |
| 'G' : { 'cmd' : function() { client.takeoff() } }, | |
| 'AE' : { 'cmd' : function() { client.up(0.5) } }, | |
| 'EA' : { 'cmd' : function() { client.down(0.5) } }, | |
| 'AGA' : { 'cmd' : function() { client.animate('flipLeft', 1000); } } | |
| } | |
| var executeSequnce = function() { | |
| if (patterns.hasOwnProperty(seq)) { | |
| var cmd = patterns[seq]; | |
| console.log("\n" + " CMD: " + seq); | |
| cmd.cmd(); | |
| } else { | |
| client.stop(); // stop motion | |
| } | |
| seq = ""; | |
| }; | |
| var delayedExecution = setInterval(executeSequnce, INTERVAL); | |
| process.stdin.on('data', function(data) { | |
| if (seq.slice(-1) !== data) { | |
| seq += data; | |
| } | |
| clearInterval(delayedExecution); | |
| delayedExecution = setInterval(executeSequnce, INTERVAL); | |
| }); |