Update app.py

app.py

@@ -11,8 +11,6 @@ from huggingface_hub import hf_hub_download
 repo_id = "Athspi/Gg"
 
 # Download the ONNX model file from the repository.
-# This will download "mms_tts_eng.onnx" from:
-# https://huggingface.co/Athspi/Gg/resolve/main/mms_tts_eng.onnx
 onnx_model_path = hf_hub_download(repo_id=repo_id, filename="mms_tts_eng.onnx")
 
 # Load the tokenizer from the repository.
@@ -34,8 +32,8 @@ def tts_inference(text: str):
         text (str): Input text to synthesize.
     
     Returns:
-        waveform (np.ndarray): Synthesized audio waveform in float32 format.
-        sampling_rate (int): The sampling rate of the waveform.
+        Tuple[int, np.ndarray]: A tuple containing the sampling rate (int) and the synthesized
+                                 audio waveform (np.ndarray in float32 format).
     """
     # Tokenize the input text.
     inputs = tokenizer(text, return_tensors="pt")
@@ -57,15 +55,21 @@ def tts_inference(text: str):
     # Remove any extra dimensions.
     waveform = np.squeeze(waveform)
     
-    return waveform, sampling_rate
+    return (sampling_rate, waveform)  # Return as a tuple
+
 
 # Build the Gradio interface.
 iface = gr.Interface(
     fn=tts_inference,
     inputs=gr.Textbox(lines=2, placeholder="Enter text here..."),
-    outputs=gr.Audio(type="numpy"),
+    outputs=gr.Audio(type="numpy", label="Generated Speech"),
     title="ONNX TTS Demo",
-    description="Text-to-Speech synthesis using an ONNX model from the Athspi/Gg repository on Hugging Face."
+    description="Text-to-Speech synthesis using an ONNX model from the Athspi/Gg repository on Hugging Face.",
+    examples=[
+        ["Hello, this is an example of text-to-speech."],
+        ["This model uses ONNX Runtime for fast inference."],
+        ["You can try your own sentences here."]
+    ]
 )
 
 if __name__ == "__main__":