Update app.py

app.py

@@ -1,6 +1,7 @@
 import gradio as gr
 from transformers import pipeline, AutoTokenizer
 import numpy as np
+from pydub import AudioSegment
 
 # Load the pipeline for speech recognition and translation
 pipe = pipeline(
@@ -12,9 +13,13 @@ translator = pipeline("text2text-generation", model="Baghdad99/saad-hausa-text-t
 tts = pipeline("text-to-speech", model="Baghdad99/english_voice_tts")
 
 # Define the function to translate speech
-def translate_speech(audio):
-    # Separate the sample rate and the audio data
-    sample_rate, audio_data = audio
+def translate_speech(audio_file):
+    # Load the audio file with pydub
+    audio = AudioSegment.from_mp3(audio_file.name)
+
+    # Convert the audio to mono and get the raw data
+    audio = audio.set_channels(1)
+    audio_data = np.array(audio.get_array_of_samples())
 
     # Use the speech recognition pipeline to transcribe the audio
     output = pipe(audio_data)
@@ -58,7 +63,7 @@ def translate_speech(audio):
 # Define the Gradio interface
 iface = gr.Interface(
     fn=translate_speech, 
-    inputs=gr.inputs.Audio(source="microphone", type="numpy"), 
+    inputs=gr.inputs.Audio(type="file"),  # Change this line
     outputs=gr.outputs.Audio(type="numpy"),
     title="Hausa to English Translation",
     description="Realtime demo for Hausa to English translation using speech recognition and text-to-speech synthesis."