osborn 0.9.113 → 0.9.115

package/dist/index.js

@@ -192,6 +192,11 @@ const livekitState = {
 // connect-with-retry loop) so the module-level HTTP server can drive them.
 let intentionalLeave = false;
 let connectRoomHook = null;
+// Mints a LISTEN-ONLY LiveKit token for the meeting canvas so it can join the
+// agent's room and play the agent's REAL TTS audio track — bit-identical to
+// the browser voice experience (2026-08-01 quality architecture; replaces the
+// synth-file chain when connected). Set from main() where the room name lives.
+let canvasTokenHook = null;
 let leaveRoomHook = null;
 // Hook for the bug-reporter skill. The /report-bug HTTP endpoint validates the
 // payload + generates the reportId in the module-level handler, then delegates
@@ -237,7 +242,7 @@ async function synthMp3(text) {
     const dgKey = process.env.DEEPGRAM_API_KEY;
     if (dgKey) {
         try {
-            const dg = await fetch('https://api.deepgram.com/v1/speak?model=aura-2-thalia-en&encoding=mp3&bit_rate=48000', {
+            const dg = await fetch('https://api.deepgram.com/v1/speak?model=aura-2-asteria-en&encoding=mp3&bit_rate=48000', {
                 method: 'POST',
                 headers: { 'Authorization': `Token ${dgKey}`, 'Content-Type': 'application/json' },
                 body: JSON.stringify({ text: text.slice(0, 4000) }),
@@ -342,6 +347,18 @@ function startApiServer(workingDir, port) {
             }
             return;
         }
+        if (req.method === 'GET' && url.pathname === '/canvas-token') {
+            // Listen-only LiveKit credentials for the meeting canvas (see hook doc).
+            const out = canvasTokenHook ? await canvasTokenHook() : null;
+            if (!out) {
+                res.writeHead(503, { 'Content-Type': 'application/json' });
+                res.end(JSON.stringify({ error: 'no active room' }));
+                return;
+            }
+            res.writeHead(200, { 'Content-Type': 'application/json' });
+            res.end(JSON.stringify(out));
+            return;
+        }
         if (req.method === 'GET' && url.pathname === '/skills') {
             // Installed skills — same list the chat's get_skills data-channel message
             // returns, exposed over HTTP so the DASHBOARD (no LiveKit connection) can
@@ -642,7 +659,7 @@ function startApiServer(workingDir, port) {
                     // WAV/linear16, not mp3: mp3 files carry encoder padding (leading/
                     // trailing silence + boundary click) — back-to-back sentence clips
                     // were heard as "cracking". WAV is gapless-safe and cheaper to decode.
-                    const dg = await fetch('https://api.deepgram.com/v1/speak?model=aura-2-thalia-en&encoding=linear16&sample_rate=48000&container=wav', {
+                    const dg = await fetch('https://api.deepgram.com/v1/speak?model=aura-2-asteria-en&encoding=linear16&sample_rate=48000&container=wav', {
                         method: 'POST',
                         headers: { 'Authorization': `Token ${dgKey}`, 'Content-Type': 'application/json' },
                         body: JSON.stringify({ text }),
@@ -1715,6 +1732,10 @@ async function main() {
     // REDIRECTED into the meeting canvas instead of suppressed — the regular
     // voice pipeline reused with the meeting as the sink (latency fix).
     let meetingAddressedUntil = 0;
+    // True while the meeting-canvas page is connected to the LiveKit room as a
+    // listener (browser-parity audio path). When set, meeting replies use the
+    // NORMAL session.say pipeline — no suppression, no redirect, no synth chain.
+    let meetingCanvasInRoom = false;
     // Distinct human speakers seen in the current meeting. In a 1:1 (one human
     // + the bot) EVERY utterance is addressed to the bot — no name needed
     // (2026-08-01: "can you open carfax..." got observer-suppressed because
@@ -1963,10 +1984,15 @@ async function main() {
                 }
             }
             else {
-                // speech_off — a natural silence boundary. Flush the buffered turns now
-                // (chunk on conversational pauses, not an arbitrary clock), unless empty.
-                if (!isBot && meetingTranscriptBuffer.length)
-                    flushMeetingBuffer(botId, false);
+                // speech_off — a natural silence boundary = the speaker finished their
+                // utterance. When the conversation latch is open (1:1 or interactive
+                // mode), that boundary IS the bot's turn — flush ADDRESSED so it
+                // replies at natural pauses, VAD-driven, no name required
+                // (user directive 2026-08-01).
+                if (!isBot && meetingTranscriptBuffer.length) {
+                    const latchOpen = Date.now() < meetingAddressedUntil || meetingSpeakers.size <= 1;
+                    flushMeetingBuffer(botId, latchOpen);
+                }
             }
         });
     }
@@ -2632,6 +2658,16 @@ async function main() {
             // re-captures in the same room → feedback). Set by PipelineDirectLLM.chat()
             // when the turn is a [MEETING —] chunk. Normal user turns are unaffected.
             if (directLLM.suppressMeetingTTS) {
+                // MEETING MODE CONDITION (user directive 2026-08-01): says route to
+                // the NATIVE Recall sink by default — no room/Chrome relay (that path
+                // was moderately-to-extremely choppy). The LiveKit parity relay stays
+                // available behind OSBORN_MEETING_AUDIO=parity for future testing.
+                if (process.env.OSBORN_MEETING_AUDIO === 'parity' && meetingCanvasInRoom && activeMeetingBotId && Date.now() < meetingAddressedUntil) {
+                    console.log(`🔊🎼 meeting reply via NATIVE session.say (canvas relays): "${data.text.slice(0, 50)}"`);
+                    markMeetingSpeaking(data.text);
+                    // no return — normal say proceeds below
+                }
+                else 
                 // ADDRESSED turn → REDIRECT the normal streaming reply into the meeting
                 // (reuse of the regular speak path; no Bash roundtrip). Silent-observer
                 // turns stay suppressed as before.
@@ -3578,6 +3614,14 @@ async function main() {
     // connect callback, not via events).
     participantConnectedHandler = async (participant) => {
         console.log(`\n👤 User joined: ${participant.identity}`);
+        // The meeting canvas joining as a listener = browser-parity audio path is
+        // LIVE: the agent's normal session.say plays through the canvas into the
+        // meeting, so tts_say suppression/redirect must stand down.
+        if (participant.identity === 'meeting-canvas') {
+            meetingCanvasInRoom = true;
+            console.log('📽️🔊 meeting-canvas joined the LiveKit room — native session.say relays to the meeting');
+            return; // not a user; skip session setup for it
+        }
         // A user (re)arriving cancels the meeting leave-grace — the bot stays.
         cancelMeetingLeaveGrace();
         // A user is present — cancel any pending agent-side "alone" leave.
@@ -5481,6 +5525,17 @@ async function main() {
     // { roomName } (which connectRoomHook resolves to). During rollout, the legacy
     // GET /room-code endpoint keeps returning the last-created room NAME so old
     // frontends still function.
+    canvasTokenHook = async () => {
+        if (!activeRoomName)
+            return null;
+        const t = new AccessToken(apiKey, apiSecret, {
+            identity: 'meeting-canvas',
+            name: 'Meeting Canvas',
+            metadata: JSON.stringify({ type: 'canvas' }),
+        });
+        t.addGrant({ roomJoin: true, room: activeRoomName, canPublish: false, canSubscribe: true, canPublishData: false });
+        return { token: await t.toJwt(), url: process.env.LIVEKIT_URL || '', room: activeRoomName };
+    };
     connectRoomHook = async () => {
         intentionalLeave = false;
         cancelIdleExitTimer();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "osborn",
-  "version": "0.9.113",
+  "version": "0.9.115",
   "description": "Voice AI coding assistant - local agent that connects to Osborn frontend",
   "type": "module",
   "bin": {