Update app.py

app.py

@@ -1,5 +1,6 @@
 #Importing all the necessary packages
 import nltk
+import soundfile 
 import librosa
 import torch
 import gradio as gr
@@ -16,8 +17,9 @@ model = Wav2Vec2ForCTC.from_pretrained(model_name, use_auth_token=token_value)
 
 def load_data(input_file):
 
+  sfo = soundfile.SoundFile(librosa.ex(input_file))
   #reading the file
-  speech, sample_rate = librosa.load(input_file)
+  speech, sample_rate = librosa.load(sfo)
   #make it 1-D
   if len(speech.shape) > 1: 
       speech = speech[:,0] + speech[:,1]