diff --git a/lib/session/call-session.js b/lib/session/call-session.js
index 11f7413a..2cfa35b1 100644
--- a/lib/session/call-session.js
+++ b/lib/session/call-session.js
@@ -1598,7 +1598,23 @@ Duration=${duration} `
   }
 
   _preCacheAudio(newTasks) {
-    for (const task of newTasks) {
+    /**
+     * only precache audio for the a queued say if we have one or more non-Config verbs
+     * ahead of it in the queue.  This is because the Config verb returns immediately
+     * and would not give us enough time to generate the audio.  The point of precaching
+     * is to take advantage of getting the audio in advance of being needed, so we need
+     * to be confident we have some time before the say verb is executed, and the Config
+     * does not give us that confidence since it returns immediately.
+     */
+    const allTasks = (this.tasks || []).concat(newTasks);
+    const skipFirst = allTasks.slice(1);
+    const idxFirstNotConfig = skipFirst.findIndex((t) => t.name !== TaskName.Config);
+    if (-1 === idxFirstNotConfig) return;
+
+    const allTasksAfterFirstNotConfig = skipFirst.slice(idxFirstNotConfig);
+
+
+    for (const task of allTasksAfterFirstNotConfig) {
       if (task.name === TaskName.Config && task.hasSynthesizer) {
         /* if they change synthesizer settings don't try to precache */
         break;