[:-7]) #inserted dummy value to eliminate inconsistency i = 0 bandpass = [800, 8500] #filter unwanted frequencies prev_time = tm.time() #initiate time waveq = queue.Queue(datacount) recdata = np.array([], dtype="float32") # for wave concatenation basename = "drone" xx = [3, 3, 3, 0, 0] # test prediction value, basic check #xx = [0,0,1,2,3,0,0,3,3,3,0,0] # test prediction value, advanced check """main code""" try: #don't want user warnings while True: data, fs = record() out = reduce_noise(data, noise) ns = fil.bandpass_filter(data, bandpass) try: p, freq, b = hmn.psddetectionresults(data) except IndexError: pass b = False b = True if b: # fs = 44100#force 44100 sample rate to prediction why? #mfcc, chroma, mel, spect, tonnetz = fex.extract_feature(data,fs)#ns changed to raw data mfcc, chroma, mel, spect, tonnetz = fex.extract_feature(ns, fs) #a,e,k = lpg.lpc(ns,10) mfcc_test = par.get_parsed_mfccdata(mfcc, chroma, mel, spect, tonnetz) #lpc_test = par.get_parsed_lpcdata(a,k,freq)
i = 0 bandpass = [600,10000]#filter unwanted frequencies prev_time= tm.time()#initiate time reccount = 0 recdata = np.array([],dtype="float32") basename = "drone" labels=[] """save server recodings in assets folder""" for root, dirs, files in os.walk("assets"): for file in files: if file.endswith(".wav"): data, fs = librosa.load(os.path.join(root, file)) tests = np.split(data, 10) for test in tests: fs = 44100 ns = fil.bandpass_filter(test,bandpass) mfcc, chroma, mel, spect, tonnetz = fex.extract_feature(ns,fs) mfcc1, chroma1, mel1, spect1, tonnetz1 = fex.extract_feature(test,fs) #a,e,k = lpg.lpc(ns,10) mfcc_test = par.get_parsed_mfccdata(mfcc, chroma,mel,spect,tonnetz) mfcc_test1 = par.get_parsed_mfccdata(mfcc1, chroma1,mel1,spect1,tonnetz1) #lpc_test = par.get_parsed_lpcdata(a,k,freq) x1 = clf.predict(mfcc_test) x11 = clf.predict(mfcc_test1) label = dist_prediction_label(int(x1)) label1 = dist_prediction_label(int(x11)) labels.append([i,file,label,label1]) i+=1 import pandas as pd df = pd.DataFrame(labels)