Update message about closed-vocab models

coqui_stt_model_manager/templates/transcribe.html

@@ -12,10 +12,11 @@
 
 {% block content %}
 <p>Transcribing with <strong>{{ model_card.name }}</strong>, created by <strong>{{ model_card.creator }}</strong>.</p>
-<p>After clicking <strong>Start Recording</strong> you can speak several sentences, the system will automatically detect silences as breaks and continue transcribing.</p>
+<p>Click <strong>Start Recording</strong> and just start talking. The system automatically detects silences as breaks and continues transcribing.</p>
 {% if "digits" in scorer_basename and "yesno" in scorer_basename %}
 <div class="bd-callout bd-callout-warning">
-Due to not having enough available data, this model is using an external scorer that only includes digits (ie. "one", "two", "three"... in {{ model_card.language }}) and the words for "yes" and "no". <strong>This means the model will not predict any other words</strong>. Try saying some digits or yes/no to try it out.
+This {{ model_card.language }} model is specialized to recognize digits (i.e. from "zero" to "nine" in {{ model_card.language }}) and the words for "yes" and "no" (if they exist in {{ model_card.language }}). This model will only predict these words.
+Specializing a model's vocabulary is commonly used in conversational STT. If you'd like to specialize this {{ model_card.language }} model to another set of words, it's as easy as <a href="https://stt.readthedocs.io/en/latest/LANGUAGE_MODEL.html">swapping out the language model</a>.
 </div>
 {% endif %}
 <div id="react-root"></div>