...

herolib/clients/whisper/convert.py

@@ -0,0 +1,107 @@
+import os
+from pydub import AudioSegment
+import whisper
+import moviepy.editor as mp
+import nltk
+from nltk.tokenize import sent_tokenize, word_tokenize
+
+# Download necessary NLTK data
+nltk.download('punkt', quiet=True)
+
+class Convertor:
+    def __init__(self, max_chars_per_part=4000,context:str = "main"):
+        self.max_chars_per_part = max_chars_per_part
+        self.context = context
+
+    @classmethod
+    def new(cls, max_chars_per_part=4000):
+        return cls(max_chars_per_part)
+
+    def process(self, path: str):
+        if path.lower().endswith(('.mp4', '.avi', '.mov')):  # Video files
+            return self.process_video(path)
+        elif path.lower().endswith(('.mp3', '.wav', '.ogg')):  # Audio files
+            return self.process_audio(path)
+        else:
+            raise ValueError("Unsupported file format")
+
+    def process_video(self, video_path: str):
+        # Extract audio from video
+        video = mp.VideoFileClip(video_path)
+        audio_path = video_path.rsplit('.', 1)[0] + '.wav'
+        video.audio.write_audiofile(audio_path)
+        video.close()
+        return audio_path
+
+    def process_audio(self, audio_path: str):
+        # Convert to WAV format if necessary
+        wav_path = audio_path.rsplit('.', 1)[0] + '.wav'
+        if not audio_path.lower().endswith('.wav'):
+            audio = AudioSegment.from_file(audio_path)
+            audio.export(wav_path, format='wav')
+        else:
+            wav_path = audio_path
+
+    def split_text(self, text):
+        parts = []
+        current_part = ""
+        paragraphs = text.split('\n\n')
+        
+        for paragraph in paragraphs:
+            sentences = sent_tokenize(paragraph)
+            for sentence in sentences:
+                if len(current_part) + len(sentence) < self.max_chars_per_part:
+                    current_part += sentence + ' '
+                else:
+                    if current_part:
+                        parts.append(current_part.strip())
+                    current_part = sentence + ' '
+            
+            # Add a paragraph break if it doesn't exceed the limit
+            if len(current_part) + 2 < self.max_chars_per_part:
+                current_part += '\n\n'
+            else:
+                parts.append(current_part.strip())
+                current_part = '\n\n'
+        
+        if current_part:
+            parts.append(current_part.strip())
+        
+        return parts
+
+    def find_natural_pause(self, text):
+        words = word_tokenize(text)
+        total_words = len(words)
+        mid_point = total_words // 2
+
+        # Look for punctuation near the middle
+        for i in range(mid_point, total_words):
+            if words[i] in '.!?':
+                return ' '.join(words[:i+1]), ' '.join(words[i+1:])
+
+        # If no punctuation found, split at the nearest space to the middle
+        return ' '.join(words[:mid_point]), ' '.join(words[mid_point:])
+    
+    def write_to_file(self, parts, output_path):
+        with open(output_path, 'w', encoding='utf-8') as f:
+            for i, part in enumerate(parts, 1):
+                f.write(f"Part {i}:\n\n")
+                f.write(part)
+                f.write("\n\n")
+                if i < len(parts):
+                    f.write("-" * 50 + "\n\n")
+    
+
+# Usage example:
+if __name__ == "__main__":
+    processor = Convertor.new() 
+    item = "/Users/despiegk1/Documents/Zoom/2024-07-16 16.42.50 Kristof De Spiegeleer's Personal Meeting Room/video1720369800.mp4"
+    transcription_parts = processor.process(item)
+    
+    processor.write_to_file(transcription_parts, output_file)    
+    
+    print(f"Transcription split into {len(transcription_parts)} parts:")
+    for i, part in enumerate(transcription_parts, 1):
+        print(f"Part {i}:")
+        print(part)
+        print("-" * 50)

herolib/clients/whisper/whisper.py

@@ -0,0 +1,118 @@
+import os
+from pydub import AudioSegment
+import whisper
+import moviepy.editor as mp
+import nltk
+from nltk.tokenize import sent_tokenize, word_tokenize
+
+# Download necessary NLTK data
+nltk.download('punkt', quiet=True)
+
+class MediaProcessor:
+    def __init__(self, max_chars_per_part=4000):
+        self.model = whisper.load_model("base.en")
+        #self.model = whisper.load_model("medium.en")
+        self.max_chars_per_part = max_chars_per_part
+
+    @classmethod
+    def new(cls, max_chars_per_part=4000):
+        return cls(max_chars_per_part)
+
+    def process(self, path: str):
+        if path.lower().endswith(('.mp4', '.avi', '.mov')):  # Video files
+            return self.process_video(path)
+        elif path.lower().endswith(('.mp3', '.wav', '.ogg')):  # Audio files
+            return self.process_audio(path)
+        else:
+            raise ValueError("Unsupported file format")
+
+    def process_video(self, video_path: str):
+        # Extract audio from video
+        video = mp.VideoFileClip(video_path)
+        audio_path = video_path.rsplit('.', 1)[0] + '.wav'
+        video.audio.write_audiofile(audio_path)
+        video.close()
+
+        # Now process the extracted audio
+        return self.process_audio(audio_path)
+
+    def process_audio(self, audio_path: str):
+        # Convert to WAV format if necessary
+        wav_path = audio_path.rsplit('.', 1)[0] + '.wav'
+        if not audio_path.lower().endswith('.wav'):
+            audio = AudioSegment.from_file(audio_path)
+            audio.export(wav_path, format='wav')
+        else:
+            wav_path = audio_path
+
+        # Transcribe audio using Whisper
+        result = self.model.transcribe(wav_path)
+        transcription = result["text"]
+
+        # Split the transcription into parts
+        return self.split_text(transcription)
+
+    def split_text(self, text):
+        parts = []
+        current_part = ""
+        paragraphs = text.split('\n\n')
+        
+        for paragraph in paragraphs:
+            sentences = sent_tokenize(paragraph)
+            for sentence in sentences:
+                if len(current_part) + len(sentence) < self.max_chars_per_part:
+                    current_part += sentence + ' '
+                else:
+                    if current_part:
+                        parts.append(current_part.strip())
+                    current_part = sentence + ' '
+            
+            # Add a paragraph break if it doesn't exceed the limit
+            if len(current_part) + 2 < self.max_chars_per_part:
+                current_part += '\n\n'
+            else:
+                parts.append(current_part.strip())
+                current_part = '\n\n'
+        
+        if current_part:
+            parts.append(current_part.strip())
+        
+        return parts
+
+    def find_natural_pause(self, text):
+        words = word_tokenize(text)
+        total_words = len(words)
+        mid_point = total_words // 2
+
+        # Look for punctuation near the middle
+        for i in range(mid_point, total_words):
+            if words[i] in '.!?':
+                return ' '.join(words[:i+1]), ' '.join(words[i+1:])
+
+        # If no punctuation found, split at the nearest space to the middle
+        return ' '.join(words[:mid_point]), ' '.join(words[mid_point:])
+    
+    def write_to_file(self, parts, output_path):
+        with open(output_path, 'w', encoding='utf-8') as f:
+            for i, part in enumerate(parts, 1):
+                f.write(f"Part {i}:\n\n")
+                f.write(part)
+                f.write("\n\n")
+                if i < len(parts):
+                    f.write("-" * 50 + "\n\n")
+    
+
+# Usage example:
+if __name__ == "__main__":
+    processor = MediaProcessor.new(max_chars_per_part=10000) 
+    output_file = "/Users/despiegk1/Documents/transcription3.md"
+    item = "/Users/despiegk1/Documents/Zoom/2024-07-16 16.42.50 Kristof De Spiegeleer's Personal Meeting Room/video1720369800.mp4"
+    transcription_parts = processor.process(item)
+    
+    processor.write_to_file(transcription_parts, output_file)    
+    
+    print(f"Transcription split into {len(transcription_parts)} parts:")
+    for i, part in enumerate(transcription_parts, 1):
+        print(f"Part {i}:")
+        print(part)
+        print("-" * 50)