@inetafrica/open-claudia 2.6.44 → 2.6.46

package/bot-agent.js

@@ -767,25 +767,9 @@ function transcribeAudio(oggPath) {
 }
 
 // ── Text-to-Speech ────────────────────────────────────────────────
+// Shared with direct mode: ElevenLabs natural voice, falling back to `say`.
 
-const TTS_CMD = process.platform === "darwin" ? "say" : null;
-
-function textToVoice(text) {
-  if (!TTS_CMD || !FFMPEG) return null;
-  try {
-    const clean = text.replace(/[*_`#>\[\]()]/g, "").replace(/\n{2,}/g, ". ").replace(/\n/g, " ").trim();
-    if (!clean) return null;
-    const aiffPath = path.join(TEMP_DIR, `tts-${Date.now()}.aiff`);
-    const oggPath = aiffPath.replace(".aiff", ".ogg");
-    execSync(`${TTS_CMD} ${JSON.stringify(clean)} -o "${aiffPath}"`, { timeout: 30000 });
-    execSync(`"${FFMPEG}" -i "${aiffPath}" -c:a libopus -y "${oggPath}" 2>/dev/null`, { timeout: 30000 });
-    try { fs.unlinkSync(aiffPath); } catch (e) {}
-    return oggPath;
-  } catch (e) {
-    console.error("TTS error:", e.message);
-    return null;
-  }
-}
+const { textToVoice } = require("./core/media");
 
 async function sendVoice(oggPath) {
   try {
@@ -1420,11 +1404,16 @@ async function runClaude(prompt, cwd, replyToMsgId, opts = {}) {
       }
       if (code !== 0 && code !== null) await send(`Exit code: ${code}`);
 
-      // Send voice reply if input was a voice note
-      if (lastInputWasVoice && TTS_CMD) {
+      // Spoken auto-replies on voice input are off by default on chat
+      // transports (they were unwanted noise on Telegram). Opt in with
+      // VOICE_REPLY_ON_VOICE=1. The hands-free voice channel speaks back
+      // regardless via its own path.
+      if (lastInputWasVoice) {
         lastInputWasVoice = false;
-        const voicePath = textToVoice(finalText);
-        if (voicePath) await sendVoice(voicePath);
+        if (process.env.VOICE_REPLY_ON_VOICE === "1") {
+          const voicePath = await textToVoice(finalText);
+          if (voicePath) await sendVoice(voicePath);
+        }
       }
     } catch (e) {
       console.error("Final message delivery failed:", e.message);

package/channels/voice/adapter.js

@@ -0,0 +1,431 @@
+// VoiceAdapter — an official Open Claudia channel for the companion app.
+//
+// Runs a LAN HTTP + WebSocket server the Expo client connects to (over
+// WireGuard / local network). Inbound: push-to-talk audio, typed text, and
+// file/image uploads. Outbound: assistant text (with live edits), spoken
+// replies (ElevenLabs ogg), and files — pushed to the client over the
+// WebSocket. Everything routes through the same core handlers as Telegram
+// and Kazee, so the voice channel has the full agent capability set.
+//
+// Security: single-owner channel. Every HTTP request and WS upgrade must
+// carry the bearer token (VOICE_BRIDGE_TOKEN). The owner identity is fixed
+// (VOICE_OWNER_USER_ID), so access.js authorizes it as the bot owner.
+
+const http = require("http");
+const fs = require("fs");
+const path = require("path");
+const crypto = require("crypto");
+const { WebSocketServer } = require("ws");
+const { TEMP_DIR, FILES_DIR } = require("../../core/config");
+const { canonicalForChannel } = require("../../core/identity");
+const { inlineKeyboardToPortable } = require("../types");
+const { parseMultipart } = require("./multipart");
+
+const MEDIA_TTL_MS = 10 * 60 * 1000; // outbound media stays fetchable 10 min
+const MAX_UPLOAD = 25 * 1024 * 1024;
+
+class VoiceAdapter {
+  constructor({ id = "voice", host, port, token, ownerUserId } = {}) {
+    this.id = id;
+    this.type = "voice";
+    this.host = host || "0.0.0.0";
+    this.port = parseInt(port, 10) || 8787;
+    this.token = token || "";
+    this.ownerUserId = String(ownerUserId || "voice-owner");
+    // Single conversation: the channel id is the owner identity.
+    this.channelId = this.ownerUserId;
+    this._listeners = { message: new Set(), action: new Set() };
+    this._server = null;
+    this._wss = null;
+    this._clients = new Set();
+    this._media = new Map(); // id -> { path, mime, fileName, expires }
+    this._commands = [];
+    this._sweepTimer = null;
+  }
+
+  on(event, fn) {
+    if (!this._listeners[event]) return () => {};
+    this._listeners[event].add(fn);
+    return () => this._listeners[event].delete(fn);
+  }
+
+  _emit(event, envelope) {
+    for (const fn of this._listeners[event] || []) {
+      try { Promise.resolve(fn(envelope)).catch((e) => console.error(`voice ${event} handler:`, e.message)); }
+      catch (e) { console.error(`voice ${event} handler:`, e.message); }
+    }
+  }
+
+  // ── lifecycle ───────────────────────────────────────────────────
+
+  async start() {
+    if (!this.token) {
+      console.error("Voice adapter: VOICE_BRIDGE_TOKEN is unset — refusing to start an unauthenticated LAN server.");
+      throw new Error("VOICE_BRIDGE_TOKEN required");
+    }
+    this._server = http.createServer((req, res) => this._handleHttp(req, res));
+    this._wss = new WebSocketServer({ noServer: true });
+    this._server.on("upgrade", (req, socket, head) => this._handleUpgrade(req, socket, head));
+    this._wss.on("connection", (ws) => this._handleWsConnection(ws));
+
+    await new Promise((resolve, reject) => {
+      this._server.once("error", reject);
+      this._server.listen(this.port, this.host, () => {
+        this._server.removeListener("error", reject);
+        resolve();
+      });
+    });
+    this._sweepTimer = setInterval(() => this._sweepMedia(), 60 * 1000);
+    if (this._sweepTimer.unref) this._sweepTimer.unref();
+    console.log(`Voice channel listening on http://${this.host}:${this.port} (ws + http)`);
+  }
+
+  async stop() {
+    if (this._sweepTimer) clearInterval(this._sweepTimer);
+    for (const ws of this._clients) { try { ws.close(); } catch (e) {} }
+    this._clients.clear();
+    try { this._wss?.close(); } catch (e) {}
+    await new Promise((resolve) => { try { this._server?.close(() => resolve()); } catch (e) { resolve(); } });
+    this._server = null;
+    this._wss = null;
+  }
+
+  _authOk(req, url) {
+    const header = req.headers["authorization"] || "";
+    if (header === `Bearer ${this.token}`) return true;
+    const qsToken = url && url.searchParams.get("token");
+    return qsToken === this.token;
+  }
+
+  // ── HTTP ────────────────────────────────────────────────────────
+
+  _handleHttp(req, res) {
+    const url = new URL(req.url, `http://${req.headers.host || "localhost"}`);
+    const pathname = url.pathname;
+
+    if (req.method === "OPTIONS") return this._cors(res, 204);
+
+    if (req.method === "GET" && (pathname === "/" || pathname === "/health")) {
+      return this._json(res, 200, { ok: true, service: "open-claudia-voice", clients: this._clients.size });
+    }
+
+    // Outbound media fetch — token via query so <audio>/download works.
+    if (req.method === "GET" && pathname.startsWith("/v1/media/")) {
+      if (!this._authOk(req, url)) return this._json(res, 401, { ok: false, error: "unauthorized" });
+      return this._serveMedia(res, pathname.slice("/v1/media/".length));
+    }
+
+    if (!this._authOk(req, url)) return this._json(res, 401, { ok: false, error: "unauthorized" });
+
+    if (req.method === "GET" && pathname === "/v1/info") {
+      return this._json(res, 200, {
+        ok: true,
+        channel: this.id,
+        owner: this.ownerUserId,
+        commands: this._commands,
+        clients: this._clients.size,
+      });
+    }
+
+    if (req.method === "POST" && pathname === "/v1/messages/text") {
+      return this._readBody(req, res, (buf) => this._onText(buf, res));
+    }
+    if (req.method === "POST" && pathname === "/v1/messages/audio") {
+      return this._readBody(req, res, (buf) => this._onUpload(buf, req, res, "voice"));
+    }
+    if (req.method === "POST" && pathname === "/v1/messages/media") {
+      return this._readBody(req, res, (buf) => this._onUpload(buf, req, res, "auto"));
+    }
+
+    return this._json(res, 404, { ok: false, error: "not found" });
+  }
+
+  _readBody(req, res, cb) {
+    const chunks = [];
+    let size = 0;
+    req.on("data", (c) => {
+      size += c.length;
+      if (size > MAX_UPLOAD) { req.destroy(); this._json(res, 413, { ok: false, error: "too large" }); return; }
+      chunks.push(c);
+    });
+    req.on("end", () => { try { cb(Buffer.concat(chunks)); } catch (e) { this._json(res, 500, { ok: false, error: e.message }); } });
+    req.on("error", () => { try { this._json(res, 400, { ok: false, error: "read error" }); } catch (e) {} });
+  }
+
+  _onText(buf, res) {
+    let body = {};
+    try { body = JSON.parse(buf.toString("utf-8") || "{}"); } catch (e) {}
+    const text = (body.text || "").toString();
+    if (!text.trim()) return this._json(res, 400, { ok: false, error: "empty text" });
+    const messageId = this._mkId("t");
+    const isCommand = text.trim().startsWith("/");
+    this._emit("message", {
+      adapter: this,
+      channelId: this.channelId,
+      canonicalUserId: canonicalForChannel("voice", this.channelId),
+      userId: this.ownerUserId,
+      type: isCommand ? "command" : "text",
+      text,
+      messageId,
+      from: { id: this.ownerUserId, name: "Owner", username: "" },
+      raw: body,
+    });
+    return this._json(res, 202, { ok: true, messageId });
+  }
+
+  _onUpload(buf, req, res, kind) {
+    const ct = req.headers["content-type"] || "";
+    const m = /boundary=("?)([^";]+)\1/i.exec(ct);
+    if (!m) return this._json(res, 400, { ok: false, error: "expected multipart/form-data" });
+    const parsed = parseMultipart(buf, m[2]);
+    const file = parsed.files[0];
+    if (!file || !file.data || !file.data.length) return this._json(res, 400, { ok: false, error: "missing file" });
+
+    const caption = parsed.fields.caption || parsed.fields.text || "";
+    let type = kind;
+    const mime = (file.contentType || "").toLowerCase();
+    if (type === "auto") {
+      if (mime.startsWith("image/")) type = "photo";
+      else if (mime.startsWith("audio/")) type = "voice";
+      else type = "document";
+    }
+    const isVoice = type === "voice";
+    const baseDir = type === "document" ? FILES_DIR : TEMP_DIR;
+    const safeName = (file.filename || `${type}-${Date.now()}`).replace(/[^\w.\-]/g, "_");
+    const ext = path.extname(safeName) || (isVoice ? ".m4a" : type === "photo" ? ".jpg" : ".bin");
+    const localPath = type === "document"
+      ? path.join(baseDir, safeName)
+      : path.join(baseDir, `voice-in-${Date.now()}${ext}`);
+    fs.writeFileSync(localPath, file.data);
+
+    const messageId = this._mkId(type);
+    const envelope = {
+      adapter: this,
+      channelId: this.channelId,
+      canonicalUserId: canonicalForChannel("voice", this.channelId),
+      userId: this.ownerUserId,
+      type,
+      text: caption,
+      caption,
+      messageId,
+      from: { id: this.ownerUserId, name: "Owner", username: "" },
+      // fileId is the already-saved local path; downloadMedia just returns it.
+      media: [{ type, fileId: localPath, fileName: file.filename || path.basename(localPath), mimeType: file.contentType, size: file.data.length }],
+      raw: {},
+    };
+    this._emit("message", envelope);
+    return this._json(res, 202, { ok: true, messageId });
+  }
+
+  // ── WebSocket ───────────────────────────────────────────────────
+
+  _handleUpgrade(req, socket, head) {
+    let url;
+    try { url = new URL(req.url, `http://${req.headers.host || "localhost"}`); }
+    catch (e) { socket.destroy(); return; }
+    if (url.pathname !== "/v1/stream" || !this._authOk(req, url)) {
+      socket.write("HTTP/1.1 401 Unauthorized\r\n\r\n");
+      socket.destroy();
+      return;
+    }
+    this._wss.handleUpgrade(req, socket, head, (ws) => this._wss.emit("connection", ws));
+  }
+
+  _handleWsConnection(ws) {
+    this._clients.add(ws);
+    ws.isAlive = true;
+    ws.on("pong", () => { ws.isAlive = true; });
+    ws.on("close", () => this._clients.delete(ws));
+    ws.on("error", () => this._clients.delete(ws));
+    ws.on("message", (data) => this._onWsMessage(data));
+    this._wsSend(ws, { kind: "hello", channel: this.id, commands: this._commands, ts: Date.now() });
+  }
+
+  // Clients may send typed messages over the socket for lowest latency.
+  _onWsMessage(data) {
+    let msg = {};
+    try { msg = JSON.parse(data.toString()); } catch (e) { return; }
+    if (msg.kind === "ping") return; // keepalive
+    if (msg.kind === "text" && (msg.text || "").trim()) {
+      const text = String(msg.text);
+      const messageId = this._mkId("t");
+      this._emit("message", {
+        adapter: this,
+        channelId: this.channelId,
+        canonicalUserId: canonicalForChannel("voice", this.channelId),
+        userId: this.ownerUserId,
+        type: text.trim().startsWith("/") ? "command" : "text",
+        text,
+        messageId,
+        from: { id: this.ownerUserId, name: "Owner", username: "" },
+        raw: msg,
+      });
+    }
+  }
+
+  _broadcast(frame) {
+    const payload = JSON.stringify(frame);
+    for (const ws of this._clients) {
+      try { if (ws.readyState === ws.OPEN) ws.send(payload); } catch (e) {}
+    }
+  }
+
+  _wsSend(ws, frame) { try { ws.send(JSON.stringify(frame)); } catch (e) {} }
+
+  // ── outbound contract (called by core/io.js) ────────────────────
+
+  _normalizeKeyboard(keyboard) {
+    if (!keyboard) return null;
+    if (keyboard.buttons) return keyboard.buttons;
+    if (keyboard.inline_keyboard) return inlineKeyboardToPortable(keyboard.inline_keyboard);
+    return null;
+  }
+
+  async send(channelId, text, opts = {}) {
+    const messageId = this._mkId("a");
+    this._broadcast({
+      kind: "message",
+      role: "assistant",
+      messageId,
+      text: text || "",
+      buttons: this._normalizeKeyboard(opts.keyboard),
+      replyTo: opts.replyTo || null,
+      ts: Date.now(),
+    });
+    return messageId;
+  }
+
+  async edit(channelId, messageId, text, opts = {}) {
+    this._broadcast({
+      kind: "edit",
+      messageId,
+      text: text || "",
+      buttons: this._normalizeKeyboard(opts.keyboard),
+      ts: Date.now(),
+    });
+  }
+
+  async delete(channelId, messageId) {
+    this._broadcast({ kind: "delete", messageId, ts: Date.now() });
+  }
+
+  async sendVoice(channelId, oggPath) {
+    try {
+      const id = this._registerMedia(oggPath, "audio/ogg", path.basename(oggPath));
+      this._broadcast({ kind: "voice", messageId: this._mkId("v"), url: `/v1/media/${id}`, mime: "audio/ogg", ts: Date.now() });
+      return true;
+    } catch (e) {
+      console.error("voice sendVoice error:", e.message);
+      return false;
+    }
+  }
+
+  async sendPhoto(channelId, filePath, caption) { return this.sendFile(channelId, filePath, caption); }
+
+  async sendFile(channelId, filePath, caption) {
+    try {
+      const fileName = path.basename(filePath);
+      const mime = this._guessMime(fileName);
+      const id = this._registerMedia(filePath, mime, fileName, /* keep */ true);
+      this._broadcast({
+        kind: "file",
+        messageId: this._mkId("f"),
+        url: `/v1/media/${id}`,
+        fileName,
+        mime,
+        caption: caption || "",
+        ts: Date.now(),
+      });
+      return true;
+    } catch (e) {
+      console.error("voice sendFile error:", e.message);
+      return false;
+    }
+  }
+
+  async typing(channelId) {
+    this._broadcast({ kind: "typing", ts: Date.now() });
+  }
+
+  // Inbound media was saved to disk at upload time; fileId holds the path.
+  async downloadMedia(media) {
+    if (!media) return null;
+    return media.fileId || null;
+  }
+
+  async registerCommands(commands) {
+    this._commands = (commands || [])
+      .filter((c) => c && c.name)
+      .map((c) => ({ name: String(c.name).replace(/^\//, ""), description: String(c.description || ""), args: typeof c.args === "string" ? c.args : "" }));
+    this._broadcast({ kind: "commands", commands: this._commands, ts: Date.now() });
+  }
+
+  // ── media store ─────────────────────────────────────────────────
+
+  // keep=true means don't delete the source after serving (files the user
+  // may still want); voice replies are throwaway and removed after TTL.
+  _registerMedia(filePath, mime, fileName, keep = false) {
+    const id = crypto.randomBytes(9).toString("hex");
+    this._media.set(id, { path: filePath, mime, fileName, keep, expires: Date.now() + MEDIA_TTL_MS });
+    return id;
+  }
+
+  _serveMedia(res, id) {
+    const entry = this._media.get(id);
+    if (!entry || !fs.existsSync(entry.path)) return this._json(res, 404, { ok: false, error: "expired" });
+    const stat = fs.statSync(entry.path);
+    res.writeHead(200, {
+      "Content-Type": entry.mime || "application/octet-stream",
+      "Content-Length": stat.size,
+      "Content-Disposition": `inline; filename="${entry.fileName || "file"}"`,
+      "Access-Control-Allow-Origin": "*",
+    });
+    fs.createReadStream(entry.path).pipe(res);
+  }
+
+  _sweepMedia() {
+    const now = Date.now();
+    for (const [id, entry] of this._media) {
+      if (entry.expires <= now) {
+        this._media.delete(id);
+        if (!entry.keep) { try { fs.unlinkSync(entry.path); } catch (e) {} }
+      }
+    }
+  }
+
+  // ── helpers ─────────────────────────────────────────────────────
+
+  _mkId(prefix) { return `${prefix}-${Date.now()}-${crypto.randomBytes(4).toString("hex")}`; }
+
+  _guessMime(fileName) {
+    const ext = path.extname(fileName).toLowerCase();
+    const map = {
+      ".png": "image/png", ".jpg": "image/jpeg", ".jpeg": "image/jpeg", ".gif": "image/gif",
+      ".webp": "image/webp", ".ogg": "audio/ogg", ".mp3": "audio/mpeg", ".m4a": "audio/mp4",
+      ".wav": "audio/wav", ".mp4": "video/mp4", ".pdf": "application/pdf", ".txt": "text/plain",
+      ".json": "application/json", ".csv": "text/csv",
+    };
+    return map[ext] || "application/octet-stream";
+  }
+
+  _json(res, code, payload) {
+    const body = JSON.stringify(payload);
+    res.writeHead(code, {
+      "Content-Type": "application/json; charset=utf-8",
+      "Content-Length": Buffer.byteLength(body),
+      "Access-Control-Allow-Origin": "*",
+    });
+    res.end(body);
+  }
+
+  _cors(res, code) {
+    res.writeHead(code, {
+      "Access-Control-Allow-Origin": "*",
+      "Access-Control-Allow-Methods": "GET, POST, OPTIONS",
+      "Access-Control-Allow-Headers": "Authorization, Content-Type",
+    });
+    res.end();
+  }
+}
+
+module.exports = { VoiceAdapter };

package/channels/voice/multipart.js

@@ -0,0 +1,70 @@
+// Minimal multipart/form-data parser. The voice channel accepts audio and
+// file uploads from the app; we parse them without pulling in a body-parser
+// dependency. Returns { fields: {name: string}, files: [{name, filename,
+// contentType, data}] }.
+
+function parseMultipart(buffer, boundary) {
+  const result = { fields: {}, files: [] };
+  if (!boundary) return result;
+  const delimiter = Buffer.from(`--${boundary}`);
+  const parts = splitBuffer(buffer, delimiter);
+  for (let part of parts) {
+    // Trim leading CRLF and ignore the closing "--" / empty preamble.
+    part = trimEdges(part);
+    if (!part.length) continue;
+    const headerEnd = indexOfBuffer(part, Buffer.from("\r\n\r\n"));
+    if (headerEnd < 0) continue;
+    const headerBlock = part.slice(0, headerEnd).toString("utf-8");
+    let data = part.slice(headerEnd + 4);
+    // Each part's body is terminated by a trailing CRLF before the next
+    // delimiter — strip it.
+    if (data.length >= 2 && data[data.length - 2] === 0x0d && data[data.length - 1] === 0x0a) {
+      data = data.slice(0, data.length - 2);
+    }
+    const disposition = /content-disposition:[^\r\n]*/i.exec(headerBlock);
+    if (!disposition) continue;
+    const nameMatch = /name="([^"]*)"/i.exec(disposition[0]);
+    const filenameMatch = /filename="([^"]*)"/i.exec(disposition[0]);
+    const ctMatch = /content-type:\s*([^\r\n]+)/i.exec(headerBlock);
+    const name = nameMatch ? nameMatch[1] : "";
+    if (filenameMatch) {
+      result.files.push({
+        name,
+        filename: filenameMatch[1],
+        contentType: ctMatch ? ctMatch[1].trim() : "application/octet-stream",
+        data,
+      });
+    } else if (name) {
+      result.fields[name] = data.toString("utf-8");
+    }
+  }
+  return result;
+}
+
+function splitBuffer(buffer, delimiter) {
+  const parts = [];
+  let start = 0;
+  let idx;
+  while ((idx = indexOfBuffer(buffer, delimiter, start)) !== -1) {
+    parts.push(buffer.slice(start, idx));
+    start = idx + delimiter.length;
+  }
+  parts.push(buffer.slice(start));
+  return parts;
+}
+
+function indexOfBuffer(haystack, needle, from = 0) {
+  return haystack.indexOf(needle, from);
+}
+
+function trimEdges(part) {
+  // Each part begins with the CRLF that followed the previous boundary.
+  // Strip one leading CRLF; closing "--\r\n" and the empty preamble fall
+  // out upstream (no header delimiter / zero length).
+  if (part.length >= 2 && part[0] === 0x0d && part[1] === 0x0a) {
+    return part.slice(2);
+  }
+  return part;
+}
+
+module.exports = { parseMultipart };

package/core/access.js

@@ -14,6 +14,9 @@ const { currentTransport, currentUserId } = require("./context");
 // /channel add can update the value without a process restart.
 function transportOwnerUserId(transport) {
   if (transport === "kazee") return config.KAZEE_OWNER_USER_ID || "";
+  // The voice channel is single-owner: the bearer token gates the connection,
+  // and every envelope carries this fixed owner id, so it authorizes as owner.
+  if (transport === "voice") return config.VOICE_OWNER_USER_ID || "voice-owner";
   return "";
 }
 

package/core/adapter-registry.js

@@ -12,6 +12,7 @@ const { setAdapters } = require("./scheduler");
 
 const { TelegramAdapter } = require("../channels/telegram/adapter");
 const { KazeeAdapter } = require("../channels/kazee/adapter");
+const { VoiceAdapter } = require("../channels/voice/adapter");
 
 const adapters = [];
 let messageHandler = null;
@@ -20,6 +21,7 @@ let actionHandler = null;
 function createAdapter(spec) {
   if (spec.type === "telegram") return new TelegramAdapter({ id: spec.id, ...spec.opts });
   if (spec.type === "kazee") return new KazeeAdapter({ id: spec.id, ...spec.opts });
+  if (spec.type === "voice") return new VoiceAdapter({ id: spec.id, ...spec.opts });
   console.error(`Unknown adapter type: ${spec.type}`);
   return null;
 }

package/core/config.js

@@ -99,6 +99,9 @@ const TRANSCRIPTS_DIR = config.TRANSCRIPTS_DIR || process.env.TRANSCRIPTS_DIR ||
 const WHISPER_CLI = config.WHISPER_CLI || "";
 const WHISPER_MODEL = config.WHISPER_MODEL || "";
 const FFMPEG = config.FFMPEG || "";
+const ELEVENLABS_API_KEY = config.ELEVENLABS_API_KEY || process.env.ELEVENLABS_API_KEY || "";
+const ELEVENLABS_VOICE_ID = config.ELEVENLABS_VOICE_ID || process.env.ELEVENLABS_VOICE_ID || "EXAVITQu4vr4xnSDxMaL";
+const ELEVENLABS_MODEL = config.ELEVENLABS_MODEL || process.env.ELEVENLABS_MODEL || "eleven_v3";
 const SOUL_FILE = config.SOUL_FILE || path.join(CONFIG_DIR, "soul.md");
 const CRONS_FILE = config.CRONS_FILE || path.join(CONFIG_DIR, "crons.json");
 const JOBS_FILE = config.JOBS_FILE || path.join(CONFIG_DIR, "jobs.json");
@@ -192,6 +195,22 @@ function loadChannels() {
         type: "kazee",
         opts: { url, token, ownerUserId, botUserId },
       });
+    } else if (type === "voice") {
+      const token = config.VOICE_BRIDGE_TOKEN;
+      if (!token) {
+        console.error(`CHANNELS includes ${entry} but VOICE_BRIDGE_TOKEN is unset — skipping.`);
+        continue;
+      }
+      channels.push({
+        id,
+        type: "voice",
+        opts: {
+          host: config.VOICE_BRIDGE_HOST || "0.0.0.0",
+          port: config.VOICE_BRIDGE_PORT || "8787",
+          token,
+          ownerUserId: config.VOICE_OWNER_USER_ID || "voice-owner",
+        },
+      });
     } else {
       console.error(`Unknown channel type: ${type} — skipping.`);
     }
@@ -233,6 +252,7 @@ module.exports = {
   TRANSCRIPT_MAX_ENTRY_CHARS,
   TRANSCRIPTS_DIR,
   WHISPER_CLI, WHISPER_MODEL, FFMPEG,
+  ELEVENLABS_API_KEY, ELEVENLABS_VOICE_ID, ELEVENLABS_MODEL,
   SOUL_FILE, CRONS_FILE, JOBS_FILE, TASKS_DIR, VAULT_FILE, AUTH_FILE, IDENTITIES_FILE,
   PEOPLE_FILE, INTROS_FILE, AUDIT_FILE,
   STATE_FILE, SESSIONS_FILE,

package/core/media.js

@@ -4,7 +4,7 @@
 const fs = require("fs");
 const path = require("path");
 const { execSync } = require("child_process");
-const { WHISPER_CLI, WHISPER_MODEL, FFMPEG, TEMP_DIR } = require("./config");
+const { WHISPER_CLI, WHISPER_MODEL, FFMPEG, TEMP_DIR, ELEVENLABS_API_KEY, ELEVENLABS_VOICE_ID, ELEVENLABS_MODEL } = require("./config");
 
 const TTS_CMD = process.platform === "darwin" ? "say" : null;
 
@@ -19,11 +19,14 @@ function transcribeAudio(oggPath) {
     .join(" ").trim();
 }
 
-function textToVoice(text) {
+function cleanForTTS(text) {
+  return text.replace(/[*_`#>\[\]()]/g, "").replace(/\n{2,}/g, ". ").replace(/\n/g, " ").trim();
+}
+
+// macOS `say` fallback. Synchronous. Returns ogg path or null.
+function sayToVoice(clean) {
   if (!TTS_CMD || !FFMPEG) return null;
   try {
-    const clean = text.replace(/[*_`#>\[\]()]/g, "").replace(/\n{2,}/g, ". ").replace(/\n/g, " ").trim();
-    if (!clean) return null;
     const aiffPath = path.join(TEMP_DIR, `tts-${Date.now()}.aiff`);
     const oggPath = aiffPath.replace(".aiff", ".ogg");
     execSync(`${TTS_CMD} ${JSON.stringify(clean)} -o "${aiffPath}"`, { timeout: 30000 });
@@ -31,9 +34,49 @@ function textToVoice(text) {
     try { fs.unlinkSync(aiffPath); } catch (e) {}
     return oggPath;
   } catch (e) {
-    console.error("TTS error:", e.message);
+    console.error("say TTS error:", e.message);
     return null;
   }
 }
 
+// Natural TTS via ElevenLabs. Returns ogg path or null on any failure.
+async function elevenLabsToVoice(clean) {
+  if (!ELEVENLABS_API_KEY || !FFMPEG) return null;
+  try {
+    const res = await fetch(`https://api.elevenlabs.io/v1/text-to-speech/${ELEVENLABS_VOICE_ID}`, {
+      method: "POST",
+      headers: { "xi-api-key": ELEVENLABS_API_KEY, "Content-Type": "application/json" },
+      body: JSON.stringify({
+        text: clean,
+        model_id: ELEVENLABS_MODEL,
+        voice_settings: { stability: 0.5, similarity_boost: 0.85, style: 0.5, use_speaker_boost: true },
+      }),
+    });
+    if (!res.ok) {
+      const body = await res.text().catch(() => "");
+      console.error(`ElevenLabs TTS failed: ${res.status} ${body}`.slice(0, 300));
+      return null;
+    }
+    const buf = Buffer.from(await res.arrayBuffer());
+    const mp3Path = path.join(TEMP_DIR, `tts-${Date.now()}.mp3`);
+    const oggPath = mp3Path.replace(".mp3", ".ogg");
+    fs.writeFileSync(mp3Path, buf);
+    execSync(`"${FFMPEG}" -i "${mp3Path}" -c:a libopus -y "${oggPath}" 2>/dev/null`, { timeout: 30000 });
+    try { fs.unlinkSync(mp3Path); } catch (e) {}
+    return oggPath;
+  } catch (e) {
+    console.error("ElevenLabs TTS error:", e.message);
+    return null;
+  }
+}
+
+// Natural voice via ElevenLabs, falling back to macOS `say` only on no-key/error.
+async function textToVoice(text) {
+  const clean = cleanForTTS(text);
+  if (!clean) return null;
+  const eleven = await elevenLabsToVoice(clean);
+  if (eleven) return eleven;
+  return sayToVoice(clean);
+}
+
 module.exports = { transcribeAudio, textToVoice, TTS_CMD };

package/core/runner.js

@@ -16,7 +16,7 @@ const { chatContext, currentChannelId, currentAdapter } = require("./context");
 const { buildSystemPrompt, promptWithDynamicContext } = require("./system-prompt");
 const { redactSensitive } = require("./redact");
 const { send, editMessage, sendVoice, splitMessage } = require("./io");
-const { textToVoice, TTS_CMD } = require("./media");
+const { textToVoice } = require("./media");
 const { killProcessTree } = require("./process-tree");
 const {
   appendProjectTranscript, transcriptProjectInfo,
@@ -1193,10 +1193,16 @@ async function runClaude(prompt, cwd, replyToMsgId, opts = {}) {
       }
       if (code !== 0 && code !== null) await send(`Exit code: ${code}`);
 
-      if (state.lastInputWasVoice && TTS_CMD) {
+      if (state.lastInputWasVoice) {
         state.lastInputWasVoice = false;
-        const voicePath = textToVoice(finalText);
-        if (voicePath) await sendVoice(voicePath);
+        // Spoken replies belong to the hands-free voice channel. On chat
+        // transports (Telegram/Kazee) an auto voice note on every voice
+        // input is unwanted noise, so gate it to the voice channel.
+        const { currentTransport } = require("./context");
+        if (currentTransport() === "voice") {
+          const voicePath = await textToVoice(finalText);
+          if (voicePath) await sendVoice(voicePath);
+        }
       }
     } catch (e) {
       console.error("Final message delivery failed:", e.message);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inetafrica/open-claudia",
-  "version": "2.6.44",
+  "version": "2.6.46",
   "description": "Your always-on AI coding assistant — Claude Code, Cursor Agent, and OpenAI Codex via Telegram or Kazee Chat",
   "main": "bot.js",
   "bin": {
@@ -61,6 +61,7 @@
   "dependencies": {
     "node-cron": "^4.2.1",
     "node-telegram-bot-api": "^0.67.0",
-    "socket.io-client": "^4.7.5"
+    "socket.io-client": "^4.7.5",
+    "ws": "^8.18.0"
   }
 }