two examples

accent_gradio.py

@@ -219,11 +219,13 @@ def record_speaker(audio):
     scipy.io.wavfile.write(original_voice_path, sr, scaled)
 
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown(""" # AccentCraft
-    ### Transform your non-native accent into a native North American accent.
-    **This is an educational app designed to transform the speech of a non-native English speaker into a native American accent.**
+    gr.Markdown(""" # AccentCoach: Transform Any Accent into American Accent.
+        **This is an educational app designed to transform the speech of a non-native English speaker into a native American accent.**
+        
+        **The tool aims to coach learners in <ins>accent reduction</ins> and pronunciation improvement. It performs much better on <ins>longer speech</ins>.**
+        **The code is based on style diffusion and adversarial training with LSLMs outlined in StyleTTS2 paper.**
 
-    **The tool aims to assist learners in <ins>accent reduction</ins> and pronunciation improvement. It performs much better on <ins>longer speech</ins>.**
+        **It is strongly advised to duplicate this space and run it on a powerful GPU. Inference time can be reduced to less than a second when utilizing an Nvidia 3090.**
     """)
     # with gr.Accordion("First-Time Users (Click Here):", open=False):
     #     gr.Markdown("""
@@ -248,35 +250,44 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
 
     with gr.Column():
         gr.Markdown("""
-    *Initiate the recording process by selecting the **Record** button. You can also upload an audio file.*  
-    """)
+            *Initiate the recording process by selecting the **Record** button. Speak Clearly and ensure a noise-free environment.*  
+            """)
         inp = gr.Audio(sources=["microphone", "upload"], format="wav", type="filepath",
-    label="Your accent:",show_download_button="True")
+            label="Original accent:",show_download_button="True")
         gr.Markdown("""
-    *Press the **Run** button to listen to your native accent:*  
-    """)
+            *Press the **Run** button to listen to your native accent:*  
+            """)
         out = gr.Audio(label="Native accent:", autoplay="True", show_download_button="True")
-    btn = gr.Button("Run")
-    btn.click(transcribe, inputs=inp, outputs=out)
-
-    gr.Markdown(
-        """
-        ## Remarks: 
-        - **The current inference may be somewhat slow due to the use of free vCPUs.**
-        - **The author is optimistic about potentially upgrading to server GPUs in the future, which would significantly
-        expedite the model's runtime to within a second.**
-        - **Longer sentences yield a more naturally flowing result.
-        Brief expressions like "Hi" or "How are you" may yield suboptimal outcomes.**
-        - **The model might occasionally produce noise or generate random speech. 
-        Consider re-recording or re-running for enhanced clarity and accuracy.**
-        - **By utilizing this application, you provide consent for your voice to
-         be synthesized by pre-trained models.**
-        - **This app has been made possible through the integration of excellent libraries such as Whisper and StyleTTS2.**
-        - **If encountering an error, please try re-running or reloading the page.**
-        - **This app primarily functions as an educational tool for English learners. 
-        The author does not endorse or support any malicious or misuse of this application.**
-        - **The user acknowledges and agrees that the use of the software is at the user's sole risk.**
-        """)
+        btn = gr.Button("Run")
+        btn.click(transcribe, inputs=inp, outputs=out)
+        gr.Examples(
+            examples=[
+                ["https://dl.sndup.net/9y9x/Albert-Einstein.wav",],
+                ["https://dl.sndup.net/p6gz/Arnold-Schwarzenegger.wav" ,], 
+            ],
+            inputs=inp,
+            outputs=out,
+            fn=transcribe,
+            cache_examples=True, 
+        )
+
+        gr.Markdown(
+            """
+            ## Remarks: 
+            - **The optimal performance of the model is achieved when running on a GPU with a 
+            minimum of 8GB of VRAM. However, due to budget constraints, the author is currently 
+            limited to utilizing the free CPU on HF, resulting in slower inference speeds.**
+            - **Longer sentences yield a more naturally flowing result.
+            Brief expressions like "Hi" or "How are you" may yield suboptimal outcomes.**
+            - **The model might occasionally produce noise or generate random speech. 
+            Consider re-recording or re-running for enhanced clarity and accuracy.**
+            - **By utilizing this application, you provide consent for your voice to
+            be synthesized by pre-trained models.**
+            - **If encountering an error, please try re-running or reloading the page.**
+            - **This app primarily functions as an educational tool for English learners. 
+            The author does not endorse or support any malicious or misuse of this application.**
+            - **The user acknowledges and agrees that the use of the software is at the user's sole risk.**
+            """)
 
 
 if __name__ == "__main__":