add sample page notes

docs/index.html

@@ -23,6 +23,7 @@ <h2>Audio Samples from Unofficial Implementation of UnivNet vocoder</h2>
 		</p></div>
         <p>Please check the Github repository for the implementation details and the pre-trained model checkpoint.</p>
         <p>The ground truth audio samples from the official demo page have a different volume from the original dataset. We believe the official ground truth audio samples have been post-processed. Therefore, the audio volume of our samples might have a different volume from the official version.</p>
+        <p><b>Notes:There was a model fix related to parameter size mismatch. The currently uploaded result is from the models before modification. It will be updated after retraining the modificated model.</b></p>
         <br>
         <h3>Synthesis for seen speakers</h3>
         <li><b>LibriTTS/train-clean-360 dataset</b> was used to both train and evaluate the model.</li>