youtube links

app.py

@@ -1,21 +1,43 @@
-import whisper
-import config
+import os
 import gradio as gr
+from utils import get_text_from_audio, get_video_from_yt
+from logger import logging
+import config
 
 
+<<<<<<< HEAD
 def get_text(audio_path):
     if not audio_path:
         return "No audio file chosen..."
     model = whisper.load_model(name='base', download_root=config.MODEL_DIR)
     results = model.transcribe(audio_path)
     return results['text']
+=======
+def get_results(video_url :str) -> str:
+
+    logging.info(f">>>Getting predictions for : {video_url}")
+
+    try :
+        video_path = get_video_from_yt(video_url=video_url, save_file_dir=config.AUDIO_FILES_DIR)
+        
+        if not video_path:
+            return "Problem while downloading the video. Please check the logs."
+        
+        text = get_text_from_audio(video_path)
+
+        if not text:
+            return "Problem generating the text. Please check the logs."
+        
+        return text
 
+    except Exception as e:
+        logging.exception(e)
+        return str(e)
+>>>>>>> 0740c74 (input through youtube added)
 
-label = gr.components.Text()
-vd = gr.components.Audio(type="filepath")
 
 
-iface = gr.Interface(fn=get_text, inputs=vd, outputs=label)
+iface = gr.Interface(fn=get_results, inputs="text", outputs="text")
 iface.launch()
 
 

config.py

@@ -1,2 +1,3 @@
 import os
 MODEL_DIR = os.path.join(os.getcwd(), 'models', 'model.h5')
+AUDIO_FILES_DIR = os.path.join(os.getcwd(), 'audio_files')

logger.py

@@ -0,0 +1,6 @@
+import logging
+
+logging.basicConfig(
+    format="[ %(asctime)s ] %(lineno)d %(name)s - %(levelname)s %(message)s",
+    level=logging.INFO
+)

requirements.txt

@@ -1,3 +1,4 @@
 openai-whisper
 gradio
-ffmpeg-python
+ffmpeg-python
+yt-dlp

utils.py

@@ -0,0 +1,59 @@
+import os
+from pathlib import Path
+from logger import logging
+from typing import Optional
+import whisper
+import config
+
+
+def get_video_from_yt(video_url : str, save_file_dir : str, video_name : str="yt_audio") -> Optional[Path]:
+    """
+    Download YouTube video as an audio file in .wav . Returns the path of the 
+    downloaded file as string
+    """
+
+    logging.info(f"Attempting youtube video download : \nURL : {video_url}")
+
+    try :
+        # create directory if not exists
+        os.makedirs(save_file_dir, exist_ok=True)
+
+        filepath = f"{save_file_dir}/{video_name}"
+
+        # download the file
+        os.system(f'yt-dlp --quiet -o {filepath} -x --audio-format "wav" {video_url}')
+        
+        logging.info(f"Download successful. \nAudio file path : {filepath}")
+
+        return f"{filepath}.wav"
+    except Exception as e:
+        logging.info("Download unsuccessful.")
+        logging.exception(e)
+        return None
+
+
+def get_text_from_audio(audio_path : str) -> Optional[str]:
+    """
+    Extracts text from audio file. 
+    """
+
+    logging.info(f"Attempting to extract text from : {audio_path}")
+
+    try :
+        model = whisper.load_model(name='base', download_root=config.MODEL_DIR)
+        results = model.transcribe(audio_path)
+
+        logging.info("Extraction successful.")
+        return results['text']
+    except Exception as e:
+        logging.info("Extraction failed.")
+        logging.exception(e)
+        return None
+
+
+
+#URL = "https://www.youtube.com/watch?v=iO5LjrQaN9s"
+#save_file_dir = "audio_files"
+#video_path  = get_video_from_yt(URL, save_file_dir)
+#if video_path:
+#   print(get_text_from_audio(video_path))