Update app.py

app.py

@@ -12,13 +12,15 @@ pipe = pipeline(
 translator = pipeline("text2text-generation", model="Baghdad99/saad-hausa-text-to-english-text")
 tts = pipeline("text-to-speech", model="Baghdad99/english_voice_tts")
 
-# Define the function to translate speech
 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
 
+    # Print the shape and type of the audio data
+    print(f"Audio data type: {type(audio_data)}, Audio data shape: {audio_data.shape}")
+
     # Use the speech recognition pipeline to transcribe the audio
     output = pipe(audio_data)
     print(f"Output: {output}")  # Print the output to see what it contains
@@ -30,6 +32,9 @@ def translate_speech(audio_data_tuple):
         print("The output does not contain 'text'")
         return
 
+    # Print the transcription
+    print(f"Transcription: {transcription}")
+
     # Use the translation pipeline to translate the transcription
     translated_text = translator(transcription, return_tensors="pt")
     print(f"Translated text: {translated_text}")  # Print the translated text to see what it contains
@@ -42,6 +47,9 @@ def translate_speech(audio_data_tuple):
         print("The translated text does not contain 'generated_token_ids'")
         return
 
+    # Print the translated text string
+    print(f"Translated text string: {translated_text_str}")
+
     # Use the text-to-speech pipeline to synthesize the translated text
     synthesised_speech = tts(translated_text_str)
     print(f"Synthesised speech: {synthesised_speech}")  # Print the synthesised speech to see what it contains
@@ -56,11 +64,15 @@ def translate_speech(audio_data_tuple):
     # Flatten the audio data
     synthesised_speech_data = synthesised_speech_data.flatten()
 
+    # Print the shape and type of the synthesised speech data
+    print(f"Synthesised speech data type: {type(synthesised_speech_data)}, Synthesised speech data shape: {synthesised_speech_data.shape}")
+
     # Scale the audio data to the range of int16 format
     synthesised_speech = (synthesised_speech_data * 32767).astype(np.int16)
 
     return 16000, synthesised_speech
 
+
 # Define the Gradio interface
 iface = gr.Interface(
     fn=translate_speech,