Update app.py

app.py

@@ -1,5 +1,7 @@
 import gradio as gr
 import moviepy.editor as mp
+import librosa
+import numpy as np
 from transformers import pipeline
 
 # Load Whisper model for speech-to-text
@@ -35,14 +37,20 @@ def generate_subtitles(video_file, language_name):
 
         # Extract audio from video using moviepy
         video = mp.VideoFileClip(video_path)
+        audio_path = "temp_audio.wav"
         audio = video.audio
-        audio.write_audiofile("temp_audio.wav", codec='pcm_s16le')
+        audio.write_audiofile(audio_path, codec='pcm_s16le')
 
         print("Starting speech-to-text transcription")
 
-        # Convert speech to text (ASR using Whisper)
-        with open("temp_audio.wav", "rb") as audio_file:
-            transcription = asr(audio_file)["text"]
+        # Load the audio file as a waveform using librosa
+        waveform, sr = librosa.load(audio_path, sr=16000)  # sr=16000 for Whisper
+
+        # Convert the waveform (numpy array) to a list of floats (required by Whisper)
+        waveform_list = waveform.tolist()
+
+        # Pass the waveform to Whisper's ASR model
+        transcription = asr(waveform_list)["text"]
 
         print("Starting translation")