add some limitations

Author: lukecq1231

funaudiochat/index.html

@@ -547,6 +547,7 @@ <h2 id="Demo-SpokenQA" style="text-align: center;">Spoken QA</h2>
 <div class="container pt-5 mt-5 shadow-lg p-5 mb-5 bg-white rounded">
 	<h2 id="Demo-Duplex" style="text-align: center;">Full-Duplex Interaction</h2>
 		<p>Fun-Audio-Chat-Duplex extends Fun-Audio-Chat to support Full-Duplex voice interaction, enabling simultaneous two-way communication where the system can listen while speaking and handle user interruptions naturally. This creates more natural and efficient human-machine voice interactions, closely mirroring human-human conversations.</p>
+		<p><b>Limitations:</b> The Full-Duplex model is still under active development and may exhibit certain limitations. In some cases, the model may generate hallucinations or produce responses that are not fully accurate. Users should be aware that the model's responses may contain errors or inconsistencies, especially in complex or ambiguous scenarios. We are continuously working to improve the model's reliability and reduce hallucination issues.</p>
 		<div class="table-responsive pt-3">
 			<table class="table table-hover pt-2" style="width: 100%; border-collapse: collapse;">
 			<thead>
@@ -612,6 +613,7 @@ <h2 id="Demo-Duplex" style="text-align: center;">Full-Duplex Interaction</h2>
 <div class="container pt-5 mt-5 shadow-lg p-5 mb-5 bg-white rounded">
 	<h2 style="text-align: center;">Disclaimer</h2>
 	<p>The content provided above is for academic purposes only and is intended to demonstrate technical capabilities. Some examples are sourced from the internet. If any content infringes on your rights, please contact us to request its removal.</p>
+	<p><b>Model Limitations:</b> Fun-Audio-Chat and Fun-Audio-Chat-Duplex are research models that may exhibit limitations and potential issues, including but not limited to hallucination problems. The models may occasionally generate inaccurate, inconsistent, or factually incorrect responses. Users should exercise caution when relying on the model's outputs, especially in critical applications. The Full-Duplex mode, in particular, is an experimental feature that may have additional limitations in handling simultaneous speech processing and context understanding. We acknowledge these limitations and are committed to ongoing improvements to enhance model reliability and reduce hallucination issues.</p>
 </div>
 
 </article>