Add application file

app.py

@@ -5,6 +5,7 @@ from pytube import YouTube
 from pprint import pprint
 from moviepy.editor import VideoFileClip
 from transformers import pipeline
+import librosa
 
 # Load the Whisper model from Hugging Face
 transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base")
@@ -38,9 +39,11 @@ def create_audio_file(video_filename):
 
 def transcribe(audio_path):
     try:
-        # Load the audio file and transcribe it using the Whisper model
-        audio_file = open(audio_path, "rb")
-        transcript = transcriber(audio_file)
+        # Load the audio file and convert it to a numpy array
+        audio, _ = librosa.load(audio_path, sr=16000)
+        
+        # Transcribe the audio using the Whisper model
+        transcript = transcriber(audio)
         return transcript["text"], ""
     except Exception as e:
         return "", str(e)

requirements.txt

@@ -2,4 +2,5 @@ gradio
 pytube
 moviepy
 transformers
-torch
+torch
+librosa