implement transcriber, downloader, data_handler

.gitignore

@@ -0,0 +1,6 @@
+.DS_Store
+.idea
+audios
+models
+transcripts
+__pycache__

.gitkeep

@@ -0,0 +1,3 @@
+sounds
+models
+transcripts

datahandler.py

@@ -0,0 +1,18 @@
+import os
+import glob
+import datetime
+
+class DataHandler:
+
+    def get_audio_files(self):
+        self.audio_files = glob.glob(os.path.join(os.path.join(os.path.dirname(__file__), 'audios'), '*.m4a'))
+
+    def remove_audio_files(self):
+        for audio_file in self.audio_files:
+            os.remove(audio_file)
+
+    def save_text_as_file(self, text):
+        timestamp = datetime.datetime.now().strftime("%Y_%m_%d_%H_%M_%S")
+        filename = f"{timestamp}.txt"
+        with open(os.path.join(os.path.join(os.path.dirname(__file__), 'transcripts'), filename), 'w') as f:
+            f.write(text)

downloader.py

@@ -0,0 +1,36 @@
+from yt_dlp import YoutubeDL
+
+
+# url_list = ['https://www.youtube.com/watch?v=sw9s-jXEaOw']
+from datahandler import DataHandler
+
+
+class VideoToAudioDownloader():
+    def __init__(self, params=None):
+        if params == None:
+            self.yt_opts = {
+                'format': 'm4a/bestaudio/best',
+                'outtmpl': 'audios/%(id)s_%(release_date)s.%(ext)s',
+                # See help(yt_dlp.postprocessor) for a list of available Postprocessors and their arguments
+                'postprocessors': [{
+                    # Extract audio using ffmpeg
+                    'key': 'FFmpegExtractAudio',
+                    'preferredcodec': 'm4a',
+                }]
+            }
+        else: self.yt_opts = params
+        self.DataHandler = DataHandler()
+
+    def download_audio_from_yt(self, url):
+        with YoutubeDL(self.yt_opts) as ydl:
+            ydl.download(url)
+
+
+"""
+
+    "progress_hooks": [yt_dlp_monitor],
+
+final_filename = None
+def yt_dlp_monitor(self, d):
+    final_filename  = d.get('info_dict').get('_filename')
+"""

main.py

@@ -0,0 +1,40 @@
+import click
+
+from downloader import VideoToAudioDownloader
+from transcriber import Transcriber
+
+
+@click.group()
+def cli():
+    pass
+
+
+@cli.command()
+@click.argument('urls', nargs=-1)
+def download(urls):
+    #click.echo(urls)
+    vad = VideoToAudioDownloader()
+    for url in urls:
+        vad.download_audio_from_yt(url)
+
+@cli.command()
+def transcribe():
+    t = Transcriber()
+    t.load_model()
+    t.DataHandler.get_audio_files()
+    for af in t.audio_files:
+        t.transcribe(af)
+    t.persist()
+    t.cleanup()
+
+if __name__ == "__main__":
+    cli()
+
+
+
+
+
+
+
+
+

transcriber.py

@@ -0,0 +1,30 @@
+import os
+import glob
+import datetime
+
+import whisper
+
+from datahandler import DataHandler
+
+
+class Transcriber:
+    def __init__(self, model_name="base"):
+        self.model_name = model_name
+        self.DataHandler = DataHandler()
+
+    def load_model(self):
+        self.model = whisper.load_model(name=self.model_name, download_root="models/")
+
+    def transcribe(self, audio_file):
+        self.transcription = self.model.transcribe(
+            audio=audio_file,
+            verbose=True,
+            fp16=False)
+        self.raw_text = self.transcription["text"]
+
+    def persist(self):
+        self.DataHandler.save_text_as_file(self.raw_text)
+
+    def cleanup(self):
+        self.DataHandler.get_audio_files()
+        self.DataHandler.remove_audio_files()