Added debug prints for video and MFCC files

Added debug print() statements for fps, framerate and length

Temporarily changed MFCC output to .txt file for debugging purposes

step2_and_step3.py

@@ -142,11 +142,17 @@ def frame_processing(img, face_cascade, eye_cascade):
     return resized
 
 
-def audio_processing(clip_num):
+def audio_processing(clip_num, frame_rate, frame_count):
+
+    print("frame_rate =", frame_rate)
+    print("frame_count =", frame_count)
+    print("estimated video length = " + str(frame_count/frame_rate) + "s")
 
     (rate, sig) = wav.read("vids/dataset/audio" + str(clip_num) + ".wav")
 
-    frame_time = 1 / 30
+    #frame_rate = 30
+    #frame_rate = 29.95904248021286
+    frame_time = 1 / frame_rate
     print("frame_time = " + str(frame_time) + "s")
 
     # Returns: A numpy array of size (NUMFRAMES by numcep) containing features. Each row holds 1 feature vector.
@@ -169,17 +175,25 @@ def data_processing(vid_num, face_cascade, eye_cascade):
     # If the input is the camera, pass 0 instead of the video file name
     #cap = cv2.VideoCapture('D:vids/dataset/video1.mp4')
     # print(vid_num)
-    print('vids/dataset/video' + str(vid_num) + '.mp4')
+    print('\nvids/dataset/video' + str(vid_num) + '.mp4')
     cap = cv2.VideoCapture('vids/dataset/video' + str(vid_num) + '.mp4')
 
+    #cap.set(cv2.CAP_PROP_FRAME_HEIGHT)
+    vid_fps = cap.get(cv2.CAP_PROP_FPS)
+    vid_total_frames = cap.get(cv2.CAP_PROP_FRAME_COUNT)
+    print("Result of cap.get(FPS) is: ", vid_fps)
+    print("Result of cap.get(FRAME_COUNT) is: ", vid_total_frames)
+
     # Check if camera opened successfully
     if (cap.isOpened() == False):
         raise Exception("Error opening video stream or file")
 
-    new_path = 'data/clip' + str(vid_num) + '/images/'
+    new_frame_path = 'data/clip' + str(vid_num) + '/images/'
+    new_audio_path = 'data/clip' + str(vid_num) + '/audio/'
 
     # Creates data folder and other subdirectories, throws error if it already exists
-    os.makedirs(new_path, exist_ok=False)
+    os.makedirs(new_frame_path, exist_ok=False)
+    os.makedirs(new_audio_path, exist_ok=True)
 
     # Read until video is completed
     i = 0
@@ -191,7 +205,7 @@ def data_processing(vid_num, face_cascade, eye_cascade):
             processed_frame = frame_processing(
                 frame, face_cascade, eye_cascade)
             cv2.imwrite(
-                (new_path + 'image' + str(i).zfill(4) + '.jpg'), processed_frame)
+                (new_frame_path + 'image' + str(i).zfill(4) + '.jpg'), processed_frame)
             i += 1
         # Break the loop
         else:
@@ -200,8 +214,10 @@ def data_processing(vid_num, face_cascade, eye_cascade):
     # When everything done, release the video capture object
     cap.release()
 
-    mfcc_array = audio_processing(vid_num)
-    np.save("data/clip" + str(vid_num) + "/mfcc", mfcc_array)
+    mfcc_array = audio_processing(vid_num, vid_fps, vid_total_frames)
+    #np.save("data/clip" + str(vid_num) + "/mfcc", mfcc_array)
+    mfcc_array_row = 0
+    np.savetxt("data/clip" + str(vid_num) + "/audio/" + str(mfcc_array_row).zfill(6) + ".txt", mfcc_array)
     print("mfcc for audio" + str(vid_num) + " saved!")
 
     return

step3_test_script_multi_mfcc.py

 from python_speech_features import mfcc
-from python_speech_features import logfbank
+#from python_speech_features import logfbank
 import scipy.io.wavfile as wav
 import numpy as np
 
-frame_rate = 30
-clip_num = 1
+#frame_rate = 24.9743
+frame_rate = 29.95904248021286
+clip_num = 2
 
 mfcc_array_row = 0
 
@@ -17,6 +18,11 @@ print("frame_time = " + str(frame_time) + "s")
 #Parameters: numcep – the number of cepstrum to return, default 13
 #For a very basic understanding, cepstrum is the information of rate of change in spectral bands
 
+# NEED 146 MFCC rows!!!
+# 30 fps gives us 177 rows
+# 25 fps gives us 148 rows for video1, that's 146 rows if we discard the top and bottom row?
+
+
 #mfcc(signal, samplerate=16000, winlen=0.025, winstep=0.01, numcep=13, nfilt=26, nfft=512, lowfreq=0, highfreq=None, preemph=0.97, ceplifter=22, appendEnergy=True, winfunc=<function <lambda>>)
 mfcc_feat = mfcc(sig, rate, 0.025, frame_time, 13, 26, 1200)
 
@@ -28,7 +34,8 @@ print(fbank_feat[1:3, :])"""
 #Save an array to a binary file in NumPy .npy format.
 #Notes: Any data saved to the file is appended to the end of the file.
 
-np.save("data/clip" + str(clip_num) + "/audio/" + str(mfcc_array_row).zfill(6), mfcc_feat)
+#np.save("data/clip" + str(clip_num) + "/audio/" + str(mfcc_array_row).zfill(6), mfcc_feat)
+np.savetxt("data/clip" + str(clip_num) + "/audio/" + str(mfcc_array_row).zfill(6) + ".txt", mfcc_feat)
 print("MFCC saved!")
 
 #maybe test using np.savetxt() to see what the mfcc array output actually is first?