Update app.py

app.py

@@ -95,7 +95,7 @@ inputs = [model, audio1, audio2]
 outputs =  gr.outputs.Audio(label="Output Audio", type='filepath')
 
 title = "FreeVC"
-description = "Gradio Demo for FreeVC: Towards High-Quality Text-Free One-Shot Voice Conversion. To use it, simply upload your audio, or click the example to load. Read more at the links below. Note: It seems that the WavLM checkpoint in HuggingFace is a little different from the one used to train FreeVC, which may degrade the performance a bit."
+description = "Gradio Demo for FreeVC: Towards High-Quality Text-Free One-Shot Voice Conversion. To use it, simply upload your audio, or click the example to load. Read more at the links below. Note: It seems that the WavLM checkpoint in HuggingFace is a little different from the one used to train FreeVC, which may degrade the performance a bit. In addition, speaker similarity can be largely affected if there are too much silence in the reference audio, so please <strong>trim</strong> it before submitting."
 article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2210.15418' target='_blank'>Paper</a> | <a href='https://github.com/OlaWod/FreeVC' target='_blank'>Github Repo</a></p>"
 
 examples=[["FreeVC", 'p225_001.wav', 'p226_002.wav'], ["FreeVC-s", 'p226_002.wav', 'p225_001.wav'], ["FreeVC (24kHz)", 'p225_001.wav', 'p226_002.wav']]