Update app.py

app.py

@@ -1,8 +1,7 @@
 import gradio as gr
+from transformers import pipeline, AutoTokenizer
 import numpy as np
 from pydub import AudioSegment
-import io
-from transformers import pipeline, AutoTokenizer
 
 # Load the pipeline for speech recognition and translation
 pipe = pipeline(
@@ -14,32 +13,22 @@ translator = pipeline("text2text-generation", model="Baghdad99/saad-hausa-text-t
 tts = pipeline("text-to-speech", model="Baghdad99/english_voice_tts")
 
 def translate_speech(audio_data_tuple):
+    print(f"Type of audio: {type(audio_data_tuple)}, Value of audio: {audio_data_tuple}")  # Debug line
+
     # Extract the audio data from the tuple
     sample_rate, audio_data = audio_data_tuple
 
-    # Convert the audio data to int16 format
-    audio_data_int16 = audio_data.astype(np.int16)
-
-    # Create an AudioSegment from the audio data
-    audio_segment = AudioSegment(
-        audio_data_int16.tobytes(),  # Audio data as bytes
-        frame_rate=sample_rate,
-        sample_width=audio_data_int16.dtype.itemsize,  # Width in bytes
-        channels=1
-    )
+    # Print the shape and type of the audio data
+    print(f"Audio data type: {type(audio_data)}, Audio data shape: {audio_data.shape}")
 
-    # Export the AudioSegment as MP3
-    mp3_buffer = io.BytesIO()
-    audio_segment.export(mp3_buffer, format="mp3")
+    # Normalize the audio data to the range [-1, 1]
+    audio_data_normalized = audio_data / np.iinfo(audio_data.dtype).max
 
-    # Now you have an MP3 file in a BytesIO buffer. You can write it to a file,
-    # send it over a network, etc. Here's how you can write it to a file:
-    with open("audio.mp3", "wb") as f:
-        f.write(mp3_buffer.getvalue())
+    # Convert the normalized audio data to float64
+    audio_data_float64 = audio_data_normalized.astype(np.float64)
 
-    # Now you can feed the MP3 file to your model
     # Use the speech recognition pipeline to transcribe the audio
-    output = pipe("audio.mp3")
+    output = pipe(audio_data_float64)
 
     print(f"Output: {output}")  # Print the output to see what it contains
 
@@ -91,6 +80,11 @@ def translate_speech(audio_data_tuple):
     return 16000, synthesised_speech
 
 
+# Define the Gradio interface
+iface = gr.Interface(
+    fn=translate_speech, 
+    inputs=gr.inputs.Audio(source="microphone"),  # Change this line""
+
 # Define the Gradio interface
 iface = gr.Interface(
     fn=translate_speech,