@voicethere/agent 0.7.5 → 0.7.6

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@voicethere/agent",
-  "version": "0.7.5",
+  "version": "0.7.6",
   "description": "VoiceThere customer agent SDK — IPC types and runtime helpers for sandboxed child bundles",
   "type": "module",
   "exports": {

package/templates/crash.ts

@@ -6,11 +6,19 @@
  * - `{ type: "crash_exit" }` → process.exit(1)
  * - `{ type: "ping", id }` → `{ type: "pong", id }`
  * - `onUserSpeechFinal` text starting with "crash" → throw
- * - other finals → speak(`echo:${text}`)
+ * - other finals → speak(`echo. ${text}`)
  * - onSessionStart → speak("ready") after 1s (voice ready waiter)
  */
 import { defineAgent, sendToClient, speak } from "@voicethere/agent";
 
+/** TTS echo prefix with a sentence boundary so Piper does not glue words. */
+export function formatEchoSpeak(text: string): string {
+  const trimmed = text.trim();
+  if (!trimmed) return "";
+  // Never glue `echo:` onto the next word; Piper skips "echo colon" on `echo:One`.
+  return `echo. ${trimmed}`;
+}
+
 export const CRASH_AGENT_MESSAGE =
   "e2e crash-agent: intentional handler failure";
 
@@ -44,7 +52,9 @@ defineAgent({
     if (/^crash\b/i.test(trimmed)) {
       throw new Error(CRASH_AGENT_MESSAGE);
     }
-    speak(sessionId, `echo:${trimmed}`);
+    const echoText = formatEchoSpeak(trimmed);
+    if (!echoText) return;
+    speak(sessionId, echoText);
   },
 
   onDataChannelMessage(ctx) {

package/templates/echo-smoke.ts

@@ -5,6 +5,14 @@
  */
 import { defineAgent, parseChatText, speak } from "@voicethere/agent";
 
+/** TTS echo prefix with a sentence boundary so Piper does not glue words. */
+export function formatEchoSpeak(text: string): string {
+  const trimmed = text.trim();
+  if (!trimmed) return "";
+  // Never glue `echo:` onto the next word; Piper skips "echo colon" on `echo:One`.
+  return `echo. ${trimmed}`;
+}
+
 defineAgent({
   onSessionStart({ sessionId }) {
     setTimeout(() => {
@@ -13,15 +21,17 @@ defineAgent({
   },
 
   onUserSpeechFinal({ sessionId, text }) {
-    const trimmed = text.trim();
-    if (!trimmed) return;
-    speak(sessionId, `echo:${trimmed}`);
+    const echoText = formatEchoSpeak(text);
+    if (!echoText) return;
+    speak(sessionId, echoText);
   },
 
   onDataChannelMessage(ctx) {
     const text = parseChatText(ctx.message);
     if (!text?.trim()) return;
     if (text.trim().toLowerCase() === "ping") return;
-    speak(ctx.sessionId, `echo:${text.trim()}`);
+    const echoText = formatEchoSpeak(text);
+    if (!echoText) return;
+    speak(ctx.sessionId, echoText);
   },
 });