@voicethere/agent 0.7.8 → 0.7.9

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@voicethere/agent",
-  "version": "0.7.8",
+  "version": "0.7.9",
   "description": "VoiceThere customer agent SDK — IPC types and runtime helpers for sandboxed child bundles",
   "type": "module",
   "exports": {
@@ -95,9 +95,9 @@
     "@node-webrtc-rust/sdk": ">=0.9.0"
   },
   "devDependencies": {
-    "@node-webrtc-rust/helpers": "0.9.0",
-    "@node-webrtc-rust/sdk": "0.9.0",
-    "@node-webrtc-rust/signaling": "0.9.0",
+    "@node-webrtc-rust/helpers": "0.9.5",
+    "@node-webrtc-rust/sdk": "0.9.5",
+    "@node-webrtc-rust/signaling": "0.9.5",
     "@types/node": "22.20.1",
     "http-server": "14.1.1",
     "ioredis": "5.11.1",

package/templates/crash.ts

@@ -6,17 +6,20 @@
  * - `{ 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(`Okay. ${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. */
+/**
+ * TTS reply prefix with a sentence boundary so Piper does not glue words.
+ * Use a word local streaming STT reliably emits a token for (`Okay.`); `echo.`
+ * measured ~5% no-token Piper renders. Never glue `echo:` onto the payload.
+ */
 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}`;
+  return `Okay. ${trimmed}`;
 }
 
 export const CRASH_AGENT_MESSAGE =

package/templates/echo-smoke.ts

@@ -5,11 +5,15 @@
  */
 import { defineAgent, parseChatText, speak } from "@voicethere/agent";
 
-/** TTS echo prefix with a sentence boundary so Piper does not glue words. */
+/**
+ * TTS reply prefix with a sentence boundary so Piper does not glue words.
+ * Use a word local streaming STT (Kroko Zipformer) reliably emits a token for;
+ * `echo.` measured ~5% no-token Piper renders (audio intact). `Okay.` does not.
+ */
 export function formatEchoSpeak(text: string): string {
   const trimmed = text.trim();
   if (!trimmed) return "";
-  return `echo. ${trimmed}`;
+  return `Okay. ${trimmed}`;
 }
 
 defineAgent({