Update app.py

app.py

@@ -18,24 +18,11 @@ def query(api_url, payload):
     return response.json()
 
 # Define the function to translate speech
-def translate_speech(audio):
-    print(f"Type of audio: {type(audio)}, Value of audio: {audio}")  # Debug line
-
-    # audio is a tuple (np.ndarray, int), we need to save it as a file
-    sample_rate, audio_data = audio
-    if isinstance(audio_data, np.ndarray) and len(audio_data.shape) == 1:  # if audio_data is 1D, reshape it to 2D
-        audio_data = np.reshape(audio_data, (-1, 1))
-    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as f:
-        sf.write(f, audio_data, sample_rate)
-        audio_file = f.name
-
-    # Convert the WAV file to MP3
-    audio_segment = AudioSegment.from_wav(audio_file)
-    mp3_file = audio_file.replace(".wav", ".mp3")
-    audio_segment.export(mp3_file, format="mp3")
+def translate_speech(audio_file):
+    print(f"Type of audio: {type(audio_file)}, Value of audio: {audio_file}")  # Debug line
 
     # Use the ASR pipeline to transcribe the audio
-    with open(mp3_file, "rb") as f:  # Change this line
+    with open(audio_file, "rb") as f:
         data = f.read()
     response = requests.post(ASR_API_URL, headers=headers, data=data)
     output = response.json()
@@ -67,7 +54,7 @@ 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.File(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."