File size: 5,483 Bytes
34b89df
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
import whisper
import os
import shutil
import cv2
from moviepy.editor import ImageSequenceClip, AudioFileClip, VideoFileClip
from tqdm import tqdm

FONT = cv2.FONT_HERSHEY_SIMPLEX
FONT_SCALE = 0.8
FONT_THICKNESS = 2

class VideoTranscriber:
    def __init__(self, model_path, video_path):
        self.model = whisper.load_model(model_path)
        self.video_path = video_path
        self.audio_path = ''
        self.text_array = []
        self.fps = 0
        self.char_width = 0

    def transcribe_video(self):
        print('Transcribing video')
        result = self.model.transcribe(self.audio_path)
        text = result["segments"][0]["text"]
        textsize = cv2.getTextSize(text, FONT, FONT_SCALE, FONT_THICKNESS)[0]
        cap = cv2.VideoCapture(self.video_path)
        width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
        height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
        asp = 16/9
        ret, frame = cap.read()
        width = frame[:, int(int(width - 1 / asp * height) / 2):width - int((width - 1 / asp * height) / 2)].shape[1]
        width = width - (width * 0.1)
        self.fps = cap.get(cv2.CAP_PROP_FPS)
        self.char_width = int(textsize[0] / len(text))
        
        for j in tqdm(result["segments"]):
            lines = []
            text = j["text"]
            end = j["end"]
            start = j["start"]
            total_frames = int((end - start) * self.fps)
            start = start * self.fps
            total_chars = len(text)
            words = text.split(" ")
            i = 0
            
            while i < len(words):
                words[i] = words[i].strip()
                if words[i] == "":
                    i += 1
                    continue
                length_in_pixels = (len(words[i]) + 1) * self.char_width
                remaining_pixels = width - length_in_pixels
                line = words[i] 
                
                while remaining_pixels > 0:
                    i += 1 
                    if i >= len(words):
                        break
                    length_in_pixels = (len(words[i]) + 1) * self.char_width
                    remaining_pixels -= length_in_pixels
                    if remaining_pixels < 0:
                        continue
                    else:
                        line += " " + words[i]
                
                line_array = [line, int(start) + 15, int(len(line) / total_chars * total_frames) + int(start) + 15]
                start = int(len(line) / total_chars * total_frames) + int(start)
                lines.append(line_array)
                self.text_array.append(line_array)
        
        cap.release()
        print('Transcription complete')
    
    def extract_audio(self):
        print('Extracting audio')
        audio_path = os.path.join(os.path.dirname(self.video_path), "audio.mp3")
        video = VideoFileClip(self.video_path)
        audio = video.audio 
        audio.write_audiofile(audio_path)
        self.audio_path = audio_path
        print('Audio extracted')
    
    def extract_frames(self, output_folder):
        print('Extracting frames')
        cap = cv2.VideoCapture(self.video_path)
        width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
        height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
        asp = width / height
        N_frames = 0
        
        while True:
            ret, frame = cap.read()
            if not ret:
                break
            
            frame = frame[:, int(int(width - 1 / asp * height) / 2):width - int((width - 1 / asp * height) / 2)]
            
            for i in self.text_array:
                if N_frames >= i[1] and N_frames <= i[2]:
                    text = i[0]
                    text_size, _ = cv2.getTextSize(text, cv2.FONT_HERSHEY_SIMPLEX, 0.8, 2)
                    text_x = int((frame.shape[1] - text_size[0]) / 2)
                    text_y = int(height/2)
                    cv2.putText(frame, text, (text_x, text_y), cv2.FONT_HERSHEY_SIMPLEX, 0.75, (0, 0, 255), 2)
                    break
            
            cv2.imwrite(os.path.join(output_folder, str(N_frames) + ".jpg"), frame)
            N_frames += 1
        
        cap.release()
        print('Frames extracted')

    def create_video(self, output_video_path):
        print('Creating video')
        image_folder = os.path.join(os.path.dirname(self.video_path), "frames")
        if not os.path.exists(image_folder):
            os.makedirs(image_folder)
        
        self.extract_frames(image_folder)
        
        images = [img for img in os.listdir(image_folder) if img.endswith(".jpg")]
        images.sort(key=lambda x: int(x.split(".")[0]))
        
        frame = cv2.imread(os.path.join(image_folder, images[0]))
        height, width, layers = frame.shape
        
        clip = ImageSequenceClip([os.path.join(image_folder, image) for image in images], fps=self.fps)
        audio = AudioFileClip(self.audio_path)
        clip = clip.set_audio(audio)
        clip.write_videofile(output_video_path)
        shutil.rmtree(image_folder)
        os.remove(os.path.join(os.path.dirname(self.video_path), "audio.mp3"))

# Example usage
model_path = "base"
# video_path = "test_videos/videoplayback.mp4"
output_video_path = "output.mp4"
# output_audio_path = "test_videos/audio.mp3"