@micdrop/server 1.0.0

package/README.md

@@ -0,0 +1,245 @@
+# @micdrop/server
+
+A Node.js library for handling real-time voice conversations with WebSocket-based audio streaming.
+
+For browser implementation, see [@micdrop/client](../client/README.md) package.
+
+## Features
+
+- 🌐 WebSocket server for real-time audio streaming
+- 🔊 Audio data handling and processing
+- 💬 Conversation state management
+- ⚡ Event-based architecture
+- 🔄 Bi-directional communication
+- 🛡️ Built-in error handling
+- 🎙️ Speech-to-text and text-to-speech integration
+- 🤖 AI conversation generation support
+- 💾 Debug mode with optional audio saving
+
+## Installation
+
+```bash
+npm install @micdrop/server
+# or
+yarn add @micdrop/server
+# or
+pnpm add @micdrop/server
+```
+
+## Quick Start
+
+```typescript
+import { WebSocketServer } from 'ws'
+import { CallSocket, CallConfig } from '@micdrop/server'
+
+// Create WebSocket server
+const wss = new WebSocketServer({ port: 8080 })
+
+// Define call configuration
+const config: CallConfig = {
+  // Initial system prompt for the conversation
+  systemPrompt: 'You are a helpful assistant',
+
+  // Optional first message from assistant
+  firstMessage: 'Hello!',
+
+  // Function to generate assistant responses
+  async generateAnswer(conversation) {
+    // Implement your LLM or response generation logic
+    return 'Assistant response'
+  },
+
+  // Function to convert speech to text
+  async speech2Text(audioBlob, lastMessagePrompt) {
+    // Implement your STT logic
+    return 'Transcribed text'
+  },
+
+  // Function to convert text to speech
+  async text2Speech(text) {
+    // Implement your TTS logic
+    return new ArrayBuffer(0) // Audio data
+  },
+
+  // Optional callback when a message is added
+  onMessage(message) {
+    console.log('New message:', message)
+  },
+
+  // Optional callback when call ends
+  onEnd(summary) {
+    console.log('Call ended:', summary)
+  },
+}
+
+// Handle new connections
+wss.on('connection', (ws) => {
+  // Create call handler with configuration
+  new CallSocket(ws, config)
+})
+```
+
+## Documentation
+
+The server package provides several core components:
+
+- **CallSocket** - Main class that handles WebSocket connections, audio streaming, and conversation flow
+- **CallConfig** - Configuration interface for customizing speech processing and conversation behavior
+- **Types** - Common TypeScript types and interfaces for messages and commands
+- **Error Handling** - Standardized error handling with specific error codes
+
+## API Reference
+
+### CallSocket
+
+The main class for managing WebSocket connections and audio streaming.
+
+```typescript
+class CallSocket {
+  constructor(socket: WebSocket, config: CallConfig)
+}
+```
+
+### CallConfig
+
+Configuration interface for customizing the call behavior.
+
+```typescript
+interface CallConfig {
+  // Initial system prompt for the conversation
+  systemPrompt: string
+
+  // Optional first message from assistant
+  firstMessage?: string
+
+  // Generate assistant's response
+  generateAnswer(conversation: Conversation): Promise<string>
+
+  // Convert audio to text
+  speech2Text(blob: Blob, prompt?: string): Promise<string>
+
+  // Convert text to audio
+  text2Speech(text: string): Promise<ArrayBuffer>
+
+  // Optional callbacks
+  onMessage?(message: ConversationMessage): void
+  onEnd?(summary: CallSummary): void
+}
+```
+
+### Message Types
+
+```typescript
+interface ConversationMessage {
+  role: 'system' | 'user' | 'assistant'
+  content: string
+}
+
+type Conversation = ConversationMessage[]
+
+interface CallSummary {
+  conversation: Conversation
+  duration: number
+}
+```
+
+## WebSocket Protocol
+
+The server implements a specific protocol for client-server communication:
+
+### Client Commands
+
+```typescript
+enum CallClientCommands {
+  StartSpeaking = 'startSpeaking',
+  StopSpeaking = 'stopSpeaking',
+  Mute = 'mute',
+}
+```
+
+### Server Commands
+
+```typescript
+enum CallServerCommands {
+  UserMessage = 'userMessage',
+  AssistantMessage = 'assistantMessage',
+  CancelLastAssistantMessage = 'cancelLastAssistantMessage',
+  EndInterview = 'endInterview',
+}
+```
+
+### Message Flow
+
+1. Client connects to WebSocket server
+2. Server sends initial assistant message (if configured)
+3. Client sends audio chunks when user speaks
+4. Server processes audio and responds with text/audio
+5. Process continues until interview ends
+
+## Error Handling
+
+The server implements standardized error handling with specific codes:
+
+```typescript
+enum CallErrorCode {
+  BadRequest = 4400,
+  Unauthorized = 4401,
+  NotFound = 4404,
+}
+```
+
+Common error scenarios:
+
+- Invalid WebSocket messages
+- Authentication failures
+- Missing or invalid parameters
+- Audio processing errors
+- Connection timeouts
+
+## Integration Example
+
+Here's an example using Fastify:
+
+```typescript
+import fastify from 'fastify'
+import fastifyWebsocket from '@fastify/websocket'
+import { CallSocket, CallConfig } from '@micdrop/server'
+
+const server = fastify()
+server.register(fastifyWebsocket)
+
+server.get('/call', { websocket: true }, (socket) => {
+  const config: CallConfig = {
+    systemPrompt: 'You are a helpful assistant',
+    // ... other config options
+  }
+  new CallSocket(socket, config)
+})
+
+server.listen({ port: 8080 })
+```
+
+## Debug Mode
+
+The server includes a debug mode that can:
+
+- Log detailed timing information
+- Save audio files for debugging (optional)
+- Track conversation state
+- Monitor WebSocket events
+
+## Browser Support
+
+The server is designed to work with any WebSocket client, but is specifically tested with:
+
+- Modern browsers supporting WebSocket API
+- Node.js clients
+- @micdrop/client package
+
+## License
+
+MIT
+
+## Author
+
+[Lonestone](https://www.lonestone.io) ([GitHub](https://github.com/lonestone))

package/dist/index.d.mts

@@ -0,0 +1,65 @@
+import WebSocket$1, { WebSocket } from 'ws';
+
+declare enum CallClientCommands {
+    StartSpeaking = "startSpeaking",
+    StopSpeaking = "stopSpeaking",
+    Mute = "mute"
+}
+declare enum CallServerCommands {
+    UserMessage = "userMessage",
+    AssistantMessage = "assistantMessage",
+    CancelLastAssistantMessage = "cancelLastAssistantMessage",
+    EndInterview = "endInterview"
+}
+interface CallConfig {
+    systemPrompt: string;
+    firstMessage?: string;
+    generateAnswer(conversation: Conversation): Promise<string>;
+    speech2Text(blob: Blob, prompt?: string): Promise<string>;
+    text2Speech(text: string): Promise<ArrayBuffer>;
+    onMessage?(message: ConversationMessage): void;
+    onEnd?(call: CallSummary): void;
+}
+interface CallSummary {
+    conversation: Conversation;
+    duration: number;
+}
+type Conversation = ConversationMessage[];
+interface ConversationMessage {
+    role: 'system' | 'user' | 'assistant';
+    content: string;
+}
+
+declare const END_INTERVIEW = "END_INTERVIEW";
+declare class CallSocket {
+    socket: WebSocket;
+    config: CallConfig;
+    private startTime;
+    private lastDebug;
+    private abortAnswer;
+    private isSpeaking;
+    private chunks;
+    private conversation;
+    constructor(socket: WebSocket, config: CallConfig);
+    private addMessage;
+    private onClose;
+    private onMessage;
+    private onStopSpeaking;
+    private answer;
+    private debug;
+}
+
+declare enum CallErrorCode {
+    BadRequest = 4400,
+    Unauthorized = 4401,
+    NotFound = 4404
+}
+declare class CallError extends Error {
+    code: number;
+    constructor(code: number, message: string);
+}
+declare function handleError(socket: WebSocket$1, error: unknown): void;
+
+declare function waitForParams<CallParams>(socket: WebSocket, validate: (params: any) => CallParams): Promise<CallParams>;
+
+export { CallClientCommands, type CallConfig, CallError, CallErrorCode, CallServerCommands, CallSocket, type CallSummary, type Conversation, type ConversationMessage, END_INTERVIEW, handleError, waitForParams };

package/dist/index.d.ts

@@ -0,0 +1,65 @@
+import WebSocket$1, { WebSocket } from 'ws';
+
+declare enum CallClientCommands {
+    StartSpeaking = "startSpeaking",
+    StopSpeaking = "stopSpeaking",
+    Mute = "mute"
+}
+declare enum CallServerCommands {
+    UserMessage = "userMessage",
+    AssistantMessage = "assistantMessage",
+    CancelLastAssistantMessage = "cancelLastAssistantMessage",
+    EndInterview = "endInterview"
+}
+interface CallConfig {
+    systemPrompt: string;
+    firstMessage?: string;
+    generateAnswer(conversation: Conversation): Promise<string>;
+    speech2Text(blob: Blob, prompt?: string): Promise<string>;
+    text2Speech(text: string): Promise<ArrayBuffer>;
+    onMessage?(message: ConversationMessage): void;
+    onEnd?(call: CallSummary): void;
+}
+interface CallSummary {
+    conversation: Conversation;
+    duration: number;
+}
+type Conversation = ConversationMessage[];
+interface ConversationMessage {
+    role: 'system' | 'user' | 'assistant';
+    content: string;
+}
+
+declare const END_INTERVIEW = "END_INTERVIEW";
+declare class CallSocket {
+    socket: WebSocket;
+    config: CallConfig;
+    private startTime;
+    private lastDebug;
+    private abortAnswer;
+    private isSpeaking;
+    private chunks;
+    private conversation;
+    constructor(socket: WebSocket, config: CallConfig);
+    private addMessage;
+    private onClose;
+    private onMessage;
+    private onStopSpeaking;
+    private answer;
+    private debug;
+}
+
+declare enum CallErrorCode {
+    BadRequest = 4400,
+    Unauthorized = 4401,
+    NotFound = 4404
+}
+declare class CallError extends Error {
+    code: number;
+    constructor(code: number, message: string);
+}
+declare function handleError(socket: WebSocket$1, error: unknown): void;
+
+declare function waitForParams<CallParams>(socket: WebSocket, validate: (params: any) => CallParams): Promise<CallParams>;
+
+export { CallClientCommands, type CallConfig, CallError, CallErrorCode, CallServerCommands, CallSocket, type CallSummary, type Conversation, type ConversationMessage, END_INTERVIEW, handleError, waitForParams };

package/dist/index.js

@@ -0,0 +1,263 @@
+"use strict";
+var __create = Object.create;
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __getProtoOf = Object.getPrototypeOf;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
+  // If the importer is in node compatibility mode or this is not an ESM
+  // file that has been converted to a CommonJS file using a Babel-
+  // compatible transform (i.e. "__esModule" has not been set), then set
+  // "default" to the CommonJS "module.exports" for node compatibility.
+  isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
+  mod
+));
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+
+// index.ts
+var index_exports = {};
+__export(index_exports, {
+  CallClientCommands: () => CallClientCommands,
+  CallError: () => CallError,
+  CallErrorCode: () => CallErrorCode,
+  CallServerCommands: () => CallServerCommands,
+  CallSocket: () => CallSocket,
+  END_INTERVIEW: () => END_INTERVIEW,
+  handleError: () => handleError,
+  waitForParams: () => waitForParams
+});
+module.exports = __toCommonJS(index_exports);
+
+// src/CallSocket.ts
+var fs = __toESM(require("fs"));
+var path = __toESM(require("path"));
+
+// src/types.ts
+var CallClientCommands = /* @__PURE__ */ ((CallClientCommands2) => {
+  CallClientCommands2["StartSpeaking"] = "startSpeaking";
+  CallClientCommands2["StopSpeaking"] = "stopSpeaking";
+  CallClientCommands2["Mute"] = "mute";
+  return CallClientCommands2;
+})(CallClientCommands || {});
+var CallServerCommands = /* @__PURE__ */ ((CallServerCommands2) => {
+  CallServerCommands2["UserMessage"] = "userMessage";
+  CallServerCommands2["AssistantMessage"] = "assistantMessage";
+  CallServerCommands2["CancelLastAssistantMessage"] = "cancelLastAssistantMessage";
+  CallServerCommands2["EndInterview"] = "endInterview";
+  return CallServerCommands2;
+})(CallServerCommands || {});
+
+// src/CallSocket.ts
+var debugEnabled = false;
+var debugSaveFile = false;
+var disableTTS = false;
+var END_INTERVIEW = "END_INTERVIEW";
+var CallSocket = class {
+  constructor(socket, config) {
+    this.socket = socket;
+    this.config = config;
+    this.startTime = Date.now();
+    this.lastDebug = Date.now();
+    // An answer can be aborted if user is speaking
+    this.abortAnswer = false;
+    // When user is speaking, we're waiting to chunks or to stop
+    this.isSpeaking = false;
+    // Chunks of user speech since user started speaking
+    this.chunks = [];
+    this.conversation = [{ role: "system", content: config.systemPrompt }];
+    this.debug(`Call started`);
+    if (config.firstMessage) {
+      this.answer(config.firstMessage);
+    } else {
+      this.config.generateAnswer(this.conversation).then((answer) => this.answer(answer)).catch((error) => {
+        console.error("[WS]", error);
+        this.socket.close();
+      });
+    }
+    this.socket.on("close", this.onClose.bind(this));
+    this.socket.on("message", this.onMessage.bind(this));
+  }
+  addMessage(message) {
+    this.conversation.push(message);
+    this.socket.send(
+      `${message.role === "user" ? "userMessage" /* UserMessage */ : "assistantMessage" /* AssistantMessage */} ${message.content}`
+    );
+    this.config.onMessage?.(message);
+  }
+  onClose() {
+    this.debug("Connection closed");
+    this.abortAnswer = true;
+    const duration = Math.round((Date.now() - this.startTime) / 1e3);
+    this.config.onEnd?.({
+      conversation: this.conversation.slice(1),
+      // Remove system message
+      duration
+    });
+    this.socket = null;
+    this.config = null;
+  }
+  async onMessage(message) {
+    if (!Buffer.isBuffer(message)) {
+      console.warn(`[WS] Message is not a buffer`);
+      return;
+    }
+    if (message.byteLength < 15) {
+      const cmd = message.toString();
+      this.debug(`Command: ${cmd}`);
+      if (cmd === "startSpeaking" /* StartSpeaking */) {
+        this.isSpeaking = true;
+        this.abortAnswer = true;
+      } else if (cmd === "mute" /* Mute */) {
+        this.abortAnswer = true;
+      } else if (cmd === "stopSpeaking" /* StopSpeaking */) {
+        this.isSpeaking = false;
+        await this.onStopSpeaking();
+      }
+    } else if (Buffer.isBuffer(message) && this.isSpeaking) {
+      this.debug(`Received chunk (${message.byteLength} bytes)`);
+      this.chunks.push(message);
+    }
+  }
+  async onStopSpeaking() {
+    if (this.chunks.length === 0) return;
+    this.abortAnswer = false;
+    const blob = new Blob(this.chunks, { type: "audio/ogg" });
+    this.chunks.length = 0;
+    try {
+      if (debugSaveFile) {
+        const filePath = path.join(__dirname, "speech.ogg");
+        fs.writeFileSync(filePath, Buffer.from(await blob.arrayBuffer()));
+        return;
+      }
+      const transcript = await this.config.speech2Text(
+        blob,
+        this.conversation[this.conversation.length - 1]?.content
+      );
+      if (!transcript) {
+        this.debug("Ignoring empty transcript");
+        return;
+      }
+      this.debug("User transcript:", transcript);
+      this.addMessage({ role: "user", content: transcript });
+      if (this.abortAnswer) {
+        this.debug("Answer aborted, no answer generated");
+        return;
+      }
+      const answer = await this.config.generateAnswer(this.conversation);
+      if (this.abortAnswer) {
+        this.debug("Answer aborted, ignoring answer");
+        return;
+      }
+      await this.answer(answer);
+    } catch (error) {
+      console.error("[WS]", error);
+      this.socket.close();
+    }
+  }
+  async answer(message) {
+    let isEnd = false;
+    if (message.includes(END_INTERVIEW)) {
+      message = message.replace(END_INTERVIEW, "").trim();
+      isEnd = true;
+    }
+    if (message.length) {
+      this.debug("Assistant message:", message);
+      this.addMessage({ role: "assistant", content: message });
+      if (!disableTTS) {
+        const audio = await this.config.text2Speech(message);
+        if (this.abortAnswer) {
+          this.debug("Answer aborted, removing last assistant message");
+          const lastMessage = this.conversation[this.conversation.length - 1];
+          if (lastMessage?.role === "assistant") {
+            this.conversation.pop();
+            this.socket.send("cancelLastAssistantMessage" /* CancelLastAssistantMessage */);
+          }
+          return;
+        }
+        this.debug(`Send audio: (${audio.byteLength} bytes)`);
+        this.socket.send(audio);
+      }
+    }
+    if (isEnd) {
+      this.debug("Interview ended");
+      this.socket.send("endInterview" /* EndInterview */);
+    }
+  }
+  debug(...message) {
+    if (!debugEnabled) return;
+    const nowTime = Date.now();
+    console.log(
+      `[WS] [${nowTime - this.startTime} | ${nowTime - this.lastDebug}ms]`,
+      ...message
+    );
+    this.lastDebug = nowTime;
+  }
+};
+
+// src/errors.ts
+var CallErrorCode = /* @__PURE__ */ ((CallErrorCode2) => {
+  CallErrorCode2[CallErrorCode2["BadRequest"] = 4400] = "BadRequest";
+  CallErrorCode2[CallErrorCode2["Unauthorized"] = 4401] = "Unauthorized";
+  CallErrorCode2[CallErrorCode2["NotFound"] = 4404] = "NotFound";
+  return CallErrorCode2;
+})(CallErrorCode || {});
+var CallError = class extends Error {
+  constructor(code, message) {
+    super(message);
+    this.code = code;
+  }
+};
+function handleError(socket, error) {
+  if (error instanceof CallError) {
+    socket.close(error.code, error.message);
+  } else {
+    console.error(error);
+    socket.close(1011);
+  }
+  socket.terminate();
+}
+
+// src/waitForParams.ts
+async function waitForParams(socket, validate) {
+  return new Promise((resolve, reject) => {
+    const timeout = setTimeout(() => {
+      reject(new CallError(4400 /* BadRequest */, "Missing params"));
+    }, 3e3);
+    const onParams = (payload) => {
+      clearTimeout(timeout);
+      socket.off("message", onParams);
+      try {
+        const params = validate(JSON.parse(payload));
+        resolve(params);
+      } catch (error) {
+        reject(new CallError(4400 /* BadRequest */, "Invalid params"));
+      }
+    };
+    socket.on("message", onParams);
+  });
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  CallClientCommands,
+  CallError,
+  CallErrorCode,
+  CallServerCommands,
+  CallSocket,
+  END_INTERVIEW,
+  handleError,
+  waitForParams
+});
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../index.ts","../src/CallSocket.ts","../src/types.ts","../src/errors.ts","../src/waitForParams.ts"],"sourcesContent":["export * from './src/CallSocket'\nexport * from './src/errors'\nexport * from './src/types'\nexport * from './src/waitForParams'\n","import * as fs from 'fs'\nimport * as path from 'path'\nimport { WebSocket } from 'ws'\nimport {\n  CallClientCommands,\n  CallConfig,\n  CallServerCommands,\n  Conversation,\n  ConversationMessage,\n} from './types'\n\nconst debugEnabled = false\nconst debugSaveFile = false\nconst disableTTS = false\n\nexport const END_INTERVIEW = 'END_INTERVIEW'\n\nexport class CallSocket {\n  private startTime = Date.now()\n  private lastDebug = Date.now()\n\n  // An answer can be aborted if user is speaking\n  private abortAnswer = false\n\n  // When user is speaking, we're waiting to chunks or to stop\n  private isSpeaking = false\n\n  // Chunks of user speech since user started speaking\n  private chunks: Buffer[] = []\n\n  // Conversation history\n  private conversation: Conversation\n\n  constructor(\n    public socket: WebSocket,\n    public config: CallConfig\n  ) {\n    this.conversation = [{ role: 'system', content: config.systemPrompt }]\n    this.debug(`Call started`)\n\n    // Assistant speaks first\n\n    // LLM: Generate answer\n    if (config.firstMessage) {\n      this.answer(config.firstMessage)\n    } else {\n      this.config\n        .generateAnswer(this.conversation)\n        .then((answer) => this.answer(answer))\n        .catch((error) => {\n          console.error('[WS]', error)\n          this.socket.close()\n        })\n    }\n\n    // Listen to events\n    this.socket.on('close', this.onClose.bind(this))\n    this.socket.on('message', this.onMessage.bind(this))\n  }\n\n  private addMessage(message: ConversationMessage) {\n    this.conversation.push(message)\n    this.socket.send(\n      `${\n        message.role === 'user'\n          ? CallServerCommands.UserMessage\n          : CallServerCommands.AssistantMessage\n      } ${message.content}`\n    )\n    this.config.onMessage?.(message)\n  }\n\n  private onClose() {\n    this.debug('Connection closed')\n    this.abortAnswer = true\n    const duration = Math.round((Date.now() - this.startTime) / 1000)\n\n    // End call callback\n    this.config.onEnd?.({\n      conversation: this.conversation.slice(1), // Remove system message\n      duration,\n    })\n\n    // Unset params\n    // @ts-ignore\n    this.socket = null\n    // @ts-ignore\n    this.config = null\n  }\n\n  private async onMessage(message: Buffer) {\n    if (!Buffer.isBuffer(message)) {\n      console.warn(`[WS] Message is not a buffer`)\n      return\n    }\n\n    // Commands\n    if (message.byteLength < 15) {\n      const cmd = message.toString()\n      this.debug(`Command: ${cmd}`)\n\n      if (cmd === CallClientCommands.StartSpeaking) {\n        // User started speaking\n        this.isSpeaking = true\n        // Abort answer if there is generation in progress\n        this.abortAnswer = true\n      } else if (cmd === CallClientCommands.Mute) {\n        // User muted the call\n        // Abort answer if there is generation in progress\n        this.abortAnswer = true\n      } else if (cmd === CallClientCommands.StopSpeaking) {\n        // User stopped speaking\n        this.isSpeaking = false\n        await this.onStopSpeaking()\n      }\n    }\n\n    // Audio chunk\n    else if (Buffer.isBuffer(message) && this.isSpeaking) {\n      this.debug(`Received chunk (${message.byteLength} bytes)`)\n      this.chunks.push(message)\n    }\n  }\n\n  private async onStopSpeaking() {\n    // Do nothing if there is no chunk\n    if (this.chunks.length === 0) return\n\n    this.abortAnswer = false\n\n    // Combine audio blob\n    const blob = new Blob(this.chunks, { type: 'audio/ogg' })\n\n    // Reset chunks for next user speech\n    this.chunks.length = 0\n\n    try {\n      // Save file to disk\n      if (debugSaveFile) {\n        const filePath = path.join(__dirname, 'speech.ogg')\n        fs.writeFileSync(filePath, Buffer.from(await blob.arrayBuffer()))\n        return\n      }\n\n      // STT: Get transcript and send to client\n      const transcript = await this.config.speech2Text(\n        blob,\n        this.conversation[this.conversation.length - 1]?.content\n      )\n      if (!transcript) {\n        this.debug('Ignoring empty transcript')\n        return\n      }\n\n      this.debug('User transcript:', transcript)\n\n      // Send transcript to client\n      this.addMessage({ role: 'user', content: transcript })\n\n      if (this.abortAnswer) {\n        this.debug('Answer aborted, no answer generated')\n        return\n      }\n\n      // LLM: Generate answer\n      const answer = await this.config.generateAnswer(this.conversation)\n      if (this.abortAnswer) {\n        this.debug('Answer aborted, ignoring answer')\n        return\n      }\n\n      await this.answer(answer)\n    } catch (error) {\n      console.error('[WS]', error)\n      this.socket.close()\n    }\n  }\n\n  private async answer(message: string) {\n    let isEnd = false\n\n    // Detect end of interview\n    if (message.includes(END_INTERVIEW)) {\n      message = message.replace(END_INTERVIEW, '').trim()\n      isEnd = true\n    }\n\n    if (message.length) {\n      // Send answer to client\n      this.debug('Assistant message:', message)\n      this.addMessage({ role: 'assistant', content: message })\n\n      // TTS: Generate answer audio\n      if (!disableTTS) {\n        const audio = await this.config.text2Speech(message)\n        if (this.abortAnswer) {\n          // Remove last assistant message if aborted\n          this.debug('Answer aborted, removing last assistant message')\n          const lastMessage = this.conversation[this.conversation.length - 1]\n          if (lastMessage?.role === 'assistant') {\n            this.conversation.pop()\n            this.socket.send(CallServerCommands.CancelLastAssistantMessage)\n          }\n          return\n        }\n\n        // Send audio to client\n        this.debug(`Send audio: (${audio.byteLength} bytes)`)\n        this.socket.send(audio)\n      }\n    }\n\n    // End of call\n    if (isEnd) {\n      this.debug('Interview ended')\n      this.socket.send(CallServerCommands.EndInterview)\n    }\n  }\n\n  private debug(...message: any[]) {\n    if (!debugEnabled) return\n    const nowTime = Date.now()\n    console.log(\n      `[WS] [${nowTime - this.startTime} | ${nowTime - this.lastDebug}ms]`,\n      ...message\n    )\n    this.lastDebug = nowTime\n  }\n}\n","export enum CallClientCommands {\n  StartSpeaking = 'startSpeaking',\n  StopSpeaking = 'stopSpeaking',\n  Mute = 'mute',\n}\n\nexport enum CallServerCommands {\n  UserMessage = 'userMessage',\n  AssistantMessage = 'assistantMessage',\n  CancelLastAssistantMessage = 'cancelLastAssistantMessage',\n  EndInterview = 'endInterview',\n}\n\nexport interface CallConfig {\n  systemPrompt: string\n  firstMessage?: string\n  generateAnswer(conversation: Conversation): Promise<string>\n  speech2Text(blob: Blob, prompt?: string): Promise<string>\n  text2Speech(text: string): Promise<ArrayBuffer>\n  onMessage?(message: ConversationMessage): void\n  onEnd?(call: CallSummary): void\n}\n\nexport interface CallSummary {\n  conversation: Conversation\n  duration: number\n}\n\nexport type Conversation = ConversationMessage[]\n\nexport interface ConversationMessage {\n  role: 'system' | 'user' | 'assistant'\n  content: string\n}\n","import WebSocket from 'ws'\n\nexport enum CallErrorCode {\n  BadRequest = 4400,\n  Unauthorized = 4401,\n  NotFound = 4404,\n}\n\nexport class CallError extends Error {\n  code: number\n\n  constructor(code: number, message: string) {\n    super(message)\n    this.code = code\n  }\n}\n\nexport function handleError(socket: WebSocket, error: unknown) {\n  if (error instanceof CallError) {\n    socket.close(error.code, error.message)\n  } else {\n    console.error(error)\n    socket.close(1011)\n  }\n  socket.terminate()\n}\n","import { WebSocket } from 'ws'\nimport { CallError, CallErrorCode } from './errors'\n\nexport async function waitForParams<CallParams>(\n  socket: WebSocket,\n  validate: (params: any) => CallParams\n): Promise<CallParams> {\n  return new Promise<CallParams>((resolve, reject) => {\n    // Handle timeout\n    const timeout = setTimeout(() => {\n      reject(new CallError(CallErrorCode.BadRequest, 'Missing params'))\n    }, 3000)\n\n    const onParams = (payload: string) => {\n      // Clear timeout and listener\n      clearTimeout(timeout)\n      socket.off('message', onParams)\n\n      try {\n        // Parse JSON payload\n        const params = validate(JSON.parse(payload))\n        resolve(params)\n      } catch (error) {\n        reject(new CallError(CallErrorCode.BadRequest, 'Invalid params'))\n      }\n    }\n\n    // Listen for params\n    socket.on('message', onParams)\n  })\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;;;ACAA,SAAoB;AACpB,WAAsB;;;ACDf,IAAK,qBAAL,kBAAKA,wBAAL;AACL,EAAAA,oBAAA,mBAAgB;AAChB,EAAAA,oBAAA,kBAAe;AACf,EAAAA,oBAAA,UAAO;AAHG,SAAAA;AAAA,GAAA;AAML,IAAK,qBAAL,kBAAKC,wBAAL;AACL,EAAAA,oBAAA,iBAAc;AACd,EAAAA,oBAAA,sBAAmB;AACnB,EAAAA,oBAAA,gCAA6B;AAC7B,EAAAA,oBAAA,kBAAe;AAJL,SAAAA;AAAA,GAAA;;;ADKZ,IAAM,eAAe;AACrB,IAAM,gBAAgB;AACtB,IAAM,aAAa;AAEZ,IAAM,gBAAgB;AAEtB,IAAM,aAAN,MAAiB;AAAA,EAgBtB,YACS,QACA,QACP;AAFO;AACA;AAjBT,SAAQ,YAAY,KAAK,IAAI;AAC7B,SAAQ,YAAY,KAAK,IAAI;AAG7B;AAAA,SAAQ,cAAc;AAGtB;AAAA,SAAQ,aAAa;AAGrB;AAAA,SAAQ,SAAmB,CAAC;AAS1B,SAAK,eAAe,CAAC,EAAE,MAAM,UAAU,SAAS,OAAO,aAAa,CAAC;AACrE,SAAK,MAAM,cAAc;AAKzB,QAAI,OAAO,cAAc;AACvB,WAAK,OAAO,OAAO,YAAY;AAAA,IACjC,OAAO;AACL,WAAK,OACF,eAAe,KAAK,YAAY,EAChC,KAAK,CAAC,WAAW,KAAK,OAAO,MAAM,CAAC,EACpC,MAAM,CAAC,UAAU;AAChB,gBAAQ,MAAM,QAAQ,KAAK;AAC3B,aAAK,OAAO,MAAM;AAAA,MACpB,CAAC;AAAA,IACL;AAGA,SAAK,OAAO,GAAG,SAAS,KAAK,QAAQ,KAAK,IAAI,CAAC;AAC/C,SAAK,OAAO,GAAG,WAAW,KAAK,UAAU,KAAK,IAAI,CAAC;AAAA,EACrD;AAAA,EAEQ,WAAW,SAA8B;AAC/C,SAAK,aAAa,KAAK,OAAO;AAC9B,SAAK,OAAO;AAAA,MACV,GACE,QAAQ,SAAS,oFAGnB,IAAI,QAAQ,OAAO;AAAA,IACrB;AACA,SAAK,OAAO,YAAY,OAAO;AAAA,EACjC;AAAA,EAEQ,UAAU;AAChB,SAAK,MAAM,mBAAmB;AAC9B,SAAK,cAAc;AACnB,UAAM,WAAW,KAAK,OAAO,KAAK,IAAI,IAAI,KAAK,aAAa,GAAI;AAGhE,SAAK,OAAO,QAAQ;AAAA,MAClB,cAAc,KAAK,aAAa,MAAM,CAAC;AAAA;AAAA,MACvC;AAAA,IACF,CAAC;AAID,SAAK,SAAS;AAEd,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,MAAc,UAAU,SAAiB;AACvC,QAAI,CAAC,OAAO,SAAS,OAAO,GAAG;AAC7B,cAAQ,KAAK,8BAA8B;AAC3C;AAAA,IACF;AAGA,QAAI,QAAQ,aAAa,IAAI;AAC3B,YAAM,MAAM,QAAQ,SAAS;AAC7B,WAAK,MAAM,YAAY,GAAG,EAAE;AAE5B,UAAI,6CAA0C;AAE5C,aAAK,aAAa;AAElB,aAAK,cAAc;AAAA,MACrB,WAAW,2BAAiC;AAG1C,aAAK,cAAc;AAAA,MACrB,WAAW,2CAAyC;AAElD,aAAK,aAAa;AAClB,cAAM,KAAK,eAAe;AAAA,MAC5B;AAAA,IACF,WAGS,OAAO,SAAS,OAAO,KAAK,KAAK,YAAY;AACpD,WAAK,MAAM,mBAAmB,QAAQ,UAAU,SAAS;AACzD,WAAK,OAAO,KAAK,OAAO;AAAA,IAC1B;AAAA,EACF;AAAA,EAEA,MAAc,iBAAiB;AAE7B,QAAI,KAAK,OAAO,WAAW,EAAG;AAE9B,SAAK,cAAc;AAGnB,UAAM,OAAO,IAAI,KAAK,KAAK,QAAQ,EAAE,MAAM,YAAY,CAAC;AAGxD,SAAK,OAAO,SAAS;AAErB,QAAI;AAEF,UAAI,eAAe;AACjB,cAAM,WAAgB,UAAK,WAAW,YAAY;AAClD,QAAG,iBAAc,UAAU,OAAO,KAAK,MAAM,KAAK,YAAY,CAAC,CAAC;AAChE;AAAA,MACF;AAGA,YAAM,aAAa,MAAM,KAAK,OAAO;AAAA,QACnC;AAAA,QACA,KAAK,aAAa,KAAK,aAAa,SAAS,CAAC,GAAG;AAAA,MACnD;AACA,UAAI,CAAC,YAAY;AACf,aAAK,MAAM,2BAA2B;AACtC;AAAA,MACF;AAEA,WAAK,MAAM,oBAAoB,UAAU;AAGzC,WAAK,WAAW,EAAE,MAAM,QAAQ,SAAS,WAAW,CAAC;AAErD,UAAI,KAAK,aAAa;AACpB,aAAK,MAAM,qCAAqC;AAChD;AAAA,MACF;AAGA,YAAM,SAAS,MAAM,KAAK,OAAO,eAAe,KAAK,YAAY;AACjE,UAAI,KAAK,aAAa;AACpB,aAAK,MAAM,iCAAiC;AAC5C;AAAA,MACF;AAEA,YAAM,KAAK,OAAO,MAAM;AAAA,IAC1B,SAAS,OAAO;AACd,cAAQ,MAAM,QAAQ,KAAK;AAC3B,WAAK,OAAO,MAAM;AAAA,IACpB;AAAA,EACF;AAAA,EAEA,MAAc,OAAO,SAAiB;AACpC,QAAI,QAAQ;AAGZ,QAAI,QAAQ,SAAS,aAAa,GAAG;AACnC,gBAAU,QAAQ,QAAQ,eAAe,EAAE,EAAE,KAAK;AAClD,cAAQ;AAAA,IACV;AAEA,QAAI,QAAQ,QAAQ;AAElB,WAAK,MAAM,sBAAsB,OAAO;AACxC,WAAK,WAAW,EAAE,MAAM,aAAa,SAAS,QAAQ,CAAC;AAGvD,UAAI,CAAC,YAAY;AACf,cAAM,QAAQ,MAAM,KAAK,OAAO,YAAY,OAAO;AACnD,YAAI,KAAK,aAAa;AAEpB,eAAK,MAAM,iDAAiD;AAC5D,gBAAM,cAAc,KAAK,aAAa,KAAK,aAAa,SAAS,CAAC;AAClE,cAAI,aAAa,SAAS,aAAa;AACrC,iBAAK,aAAa,IAAI;AACtB,iBAAK,OAAO,kEAAkD;AAAA,UAChE;AACA;AAAA,QACF;AAGA,aAAK,MAAM,gBAAgB,MAAM,UAAU,SAAS;AACpD,aAAK,OAAO,KAAK,KAAK;AAAA,MACxB;AAAA,IACF;AAGA,QAAI,OAAO;AACT,WAAK,MAAM,iBAAiB;AAC5B,WAAK,OAAO,sCAAoC;AAAA,IAClD;AAAA,EACF;AAAA,EAEQ,SAAS,SAAgB;AAC/B,QAAI,CAAC,aAAc;AACnB,UAAM,UAAU,KAAK,IAAI;AACzB,YAAQ;AAAA,MACN,SAAS,UAAU,KAAK,SAAS,MAAM,UAAU,KAAK,SAAS;AAAA,MAC/D,GAAG;AAAA,IACL;AACA,SAAK,YAAY;AAAA,EACnB;AACF;;;AElOO,IAAK,gBAAL,kBAAKC,mBAAL;AACL,EAAAA,8BAAA,gBAAa,QAAb;AACA,EAAAA,8BAAA,kBAAe,QAAf;AACA,EAAAA,8BAAA,cAAW,QAAX;AAHU,SAAAA;AAAA,GAAA;AAML,IAAM,YAAN,cAAwB,MAAM;AAAA,EAGnC,YAAY,MAAc,SAAiB;AACzC,UAAM,OAAO;AACb,SAAK,OAAO;AAAA,EACd;AACF;AAEO,SAAS,YAAY,QAAmB,OAAgB;AAC7D,MAAI,iBAAiB,WAAW;AAC9B,WAAO,MAAM,MAAM,MAAM,MAAM,OAAO;AAAA,EACxC,OAAO;AACL,YAAQ,MAAM,KAAK;AACnB,WAAO,MAAM,IAAI;AAAA,EACnB;AACA,SAAO,UAAU;AACnB;;;ACtBA,eAAsB,cACpB,QACA,UACqB;AACrB,SAAO,IAAI,QAAoB,CAAC,SAAS,WAAW;AAElD,UAAM,UAAU,WAAW,MAAM;AAC/B,aAAO,IAAI,iCAAoC,gBAAgB,CAAC;AAAA,IAClE,GAAG,GAAI;AAEP,UAAM,WAAW,CAAC,YAAoB;AAEpC,mBAAa,OAAO;AACpB,aAAO,IAAI,WAAW,QAAQ;AAE9B,UAAI;AAEF,cAAM,SAAS,SAAS,KAAK,MAAM,OAAO,CAAC;AAC3C,gBAAQ,MAAM;AAAA,MAChB,SAAS,OAAO;AACd,eAAO,IAAI,iCAAoC,gBAAgB,CAAC;AAAA,MAClE;AAAA,IACF;AAGA,WAAO,GAAG,WAAW,QAAQ;AAAA,EAC/B,CAAC;AACH;","names":["CallClientCommands","CallServerCommands","CallErrorCode"]}

package/dist/index.mjs

@@ -0,0 +1,219 @@
+// src/CallSocket.ts
+import * as fs from "fs";
+import * as path from "path";
+
+// src/types.ts
+var CallClientCommands = /* @__PURE__ */ ((CallClientCommands2) => {
+  CallClientCommands2["StartSpeaking"] = "startSpeaking";
+  CallClientCommands2["StopSpeaking"] = "stopSpeaking";
+  CallClientCommands2["Mute"] = "mute";
+  return CallClientCommands2;
+})(CallClientCommands || {});
+var CallServerCommands = /* @__PURE__ */ ((CallServerCommands2) => {
+  CallServerCommands2["UserMessage"] = "userMessage";
+  CallServerCommands2["AssistantMessage"] = "assistantMessage";
+  CallServerCommands2["CancelLastAssistantMessage"] = "cancelLastAssistantMessage";
+  CallServerCommands2["EndInterview"] = "endInterview";
+  return CallServerCommands2;
+})(CallServerCommands || {});
+
+// src/CallSocket.ts
+var debugEnabled = false;
+var debugSaveFile = false;
+var disableTTS = false;
+var END_INTERVIEW = "END_INTERVIEW";
+var CallSocket = class {
+  constructor(socket, config) {
+    this.socket = socket;
+    this.config = config;
+    this.startTime = Date.now();
+    this.lastDebug = Date.now();
+    // An answer can be aborted if user is speaking
+    this.abortAnswer = false;
+    // When user is speaking, we're waiting to chunks or to stop
+    this.isSpeaking = false;
+    // Chunks of user speech since user started speaking
+    this.chunks = [];
+    this.conversation = [{ role: "system", content: config.systemPrompt }];
+    this.debug(`Call started`);
+    if (config.firstMessage) {
+      this.answer(config.firstMessage);
+    } else {
+      this.config.generateAnswer(this.conversation).then((answer) => this.answer(answer)).catch((error) => {
+        console.error("[WS]", error);
+        this.socket.close();
+      });
+    }
+    this.socket.on("close", this.onClose.bind(this));
+    this.socket.on("message", this.onMessage.bind(this));
+  }
+  addMessage(message) {
+    this.conversation.push(message);
+    this.socket.send(
+      `${message.role === "user" ? "userMessage" /* UserMessage */ : "assistantMessage" /* AssistantMessage */} ${message.content}`
+    );
+    this.config.onMessage?.(message);
+  }
+  onClose() {
+    this.debug("Connection closed");
+    this.abortAnswer = true;
+    const duration = Math.round((Date.now() - this.startTime) / 1e3);
+    this.config.onEnd?.({
+      conversation: this.conversation.slice(1),
+      // Remove system message
+      duration
+    });
+    this.socket = null;
+    this.config = null;
+  }
+  async onMessage(message) {
+    if (!Buffer.isBuffer(message)) {
+      console.warn(`[WS] Message is not a buffer`);
+      return;
+    }
+    if (message.byteLength < 15) {
+      const cmd = message.toString();
+      this.debug(`Command: ${cmd}`);
+      if (cmd === "startSpeaking" /* StartSpeaking */) {
+        this.isSpeaking = true;
+        this.abortAnswer = true;
+      } else if (cmd === "mute" /* Mute */) {
+        this.abortAnswer = true;
+      } else if (cmd === "stopSpeaking" /* StopSpeaking */) {
+        this.isSpeaking = false;
+        await this.onStopSpeaking();
+      }
+    } else if (Buffer.isBuffer(message) && this.isSpeaking) {
+      this.debug(`Received chunk (${message.byteLength} bytes)`);
+      this.chunks.push(message);
+    }
+  }
+  async onStopSpeaking() {
+    if (this.chunks.length === 0) return;
+    this.abortAnswer = false;
+    const blob = new Blob(this.chunks, { type: "audio/ogg" });
+    this.chunks.length = 0;
+    try {
+      if (debugSaveFile) {
+        const filePath = path.join(__dirname, "speech.ogg");
+        fs.writeFileSync(filePath, Buffer.from(await blob.arrayBuffer()));
+        return;
+      }
+      const transcript = await this.config.speech2Text(
+        blob,
+        this.conversation[this.conversation.length - 1]?.content
+      );
+      if (!transcript) {
+        this.debug("Ignoring empty transcript");
+        return;
+      }
+      this.debug("User transcript:", transcript);
+      this.addMessage({ role: "user", content: transcript });
+      if (this.abortAnswer) {
+        this.debug("Answer aborted, no answer generated");
+        return;
+      }
+      const answer = await this.config.generateAnswer(this.conversation);
+      if (this.abortAnswer) {
+        this.debug("Answer aborted, ignoring answer");
+        return;
+      }
+      await this.answer(answer);
+    } catch (error) {
+      console.error("[WS]", error);
+      this.socket.close();
+    }
+  }
+  async answer(message) {
+    let isEnd = false;
+    if (message.includes(END_INTERVIEW)) {
+      message = message.replace(END_INTERVIEW, "").trim();
+      isEnd = true;
+    }
+    if (message.length) {
+      this.debug("Assistant message:", message);
+      this.addMessage({ role: "assistant", content: message });
+      if (!disableTTS) {
+        const audio = await this.config.text2Speech(message);
+        if (this.abortAnswer) {
+          this.debug("Answer aborted, removing last assistant message");
+          const lastMessage = this.conversation[this.conversation.length - 1];
+          if (lastMessage?.role === "assistant") {
+            this.conversation.pop();
+            this.socket.send("cancelLastAssistantMessage" /* CancelLastAssistantMessage */);
+          }
+          return;
+        }
+        this.debug(`Send audio: (${audio.byteLength} bytes)`);
+        this.socket.send(audio);
+      }
+    }
+    if (isEnd) {
+      this.debug("Interview ended");
+      this.socket.send("endInterview" /* EndInterview */);
+    }
+  }
+  debug(...message) {
+    if (!debugEnabled) return;
+    const nowTime = Date.now();
+    console.log(
+      `[WS] [${nowTime - this.startTime} | ${nowTime - this.lastDebug}ms]`,
+      ...message
+    );
+    this.lastDebug = nowTime;
+  }
+};
+
+// src/errors.ts
+var CallErrorCode = /* @__PURE__ */ ((CallErrorCode2) => {
+  CallErrorCode2[CallErrorCode2["BadRequest"] = 4400] = "BadRequest";
+  CallErrorCode2[CallErrorCode2["Unauthorized"] = 4401] = "Unauthorized";
+  CallErrorCode2[CallErrorCode2["NotFound"] = 4404] = "NotFound";
+  return CallErrorCode2;
+})(CallErrorCode || {});
+var CallError = class extends Error {
+  constructor(code, message) {
+    super(message);
+    this.code = code;
+  }
+};
+function handleError(socket, error) {
+  if (error instanceof CallError) {
+    socket.close(error.code, error.message);
+  } else {
+    console.error(error);
+    socket.close(1011);
+  }
+  socket.terminate();
+}
+
+// src/waitForParams.ts
+async function waitForParams(socket, validate) {
+  return new Promise((resolve, reject) => {
+    const timeout = setTimeout(() => {
+      reject(new CallError(4400 /* BadRequest */, "Missing params"));
+    }, 3e3);
+    const onParams = (payload) => {
+      clearTimeout(timeout);
+      socket.off("message", onParams);
+      try {
+        const params = validate(JSON.parse(payload));
+        resolve(params);
+      } catch (error) {
+        reject(new CallError(4400 /* BadRequest */, "Invalid params"));
+      }
+    };
+    socket.on("message", onParams);
+  });
+}
+export {
+  CallClientCommands,
+  CallError,
+  CallErrorCode,
+  CallServerCommands,
+  CallSocket,
+  END_INTERVIEW,
+  handleError,
+  waitForParams
+};
+//# sourceMappingURL=index.mjs.map

package/dist/index.mjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/CallSocket.ts","../src/types.ts","../src/errors.ts","../src/waitForParams.ts"],"sourcesContent":["import * as fs from 'fs'\nimport * as path from 'path'\nimport { WebSocket } from 'ws'\nimport {\n  CallClientCommands,\n  CallConfig,\n  CallServerCommands,\n  Conversation,\n  ConversationMessage,\n} from './types'\n\nconst debugEnabled = false\nconst debugSaveFile = false\nconst disableTTS = false\n\nexport const END_INTERVIEW = 'END_INTERVIEW'\n\nexport class CallSocket {\n  private startTime = Date.now()\n  private lastDebug = Date.now()\n\n  // An answer can be aborted if user is speaking\n  private abortAnswer = false\n\n  // When user is speaking, we're waiting to chunks or to stop\n  private isSpeaking = false\n\n  // Chunks of user speech since user started speaking\n  private chunks: Buffer[] = []\n\n  // Conversation history\n  private conversation: Conversation\n\n  constructor(\n    public socket: WebSocket,\n    public config: CallConfig\n  ) {\n    this.conversation = [{ role: 'system', content: config.systemPrompt }]\n    this.debug(`Call started`)\n\n    // Assistant speaks first\n\n    // LLM: Generate answer\n    if (config.firstMessage) {\n      this.answer(config.firstMessage)\n    } else {\n      this.config\n        .generateAnswer(this.conversation)\n        .then((answer) => this.answer(answer))\n        .catch((error) => {\n          console.error('[WS]', error)\n          this.socket.close()\n        })\n    }\n\n    // Listen to events\n    this.socket.on('close', this.onClose.bind(this))\n    this.socket.on('message', this.onMessage.bind(this))\n  }\n\n  private addMessage(message: ConversationMessage) {\n    this.conversation.push(message)\n    this.socket.send(\n      `${\n        message.role === 'user'\n          ? CallServerCommands.UserMessage\n          : CallServerCommands.AssistantMessage\n      } ${message.content}`\n    )\n    this.config.onMessage?.(message)\n  }\n\n  private onClose() {\n    this.debug('Connection closed')\n    this.abortAnswer = true\n    const duration = Math.round((Date.now() - this.startTime) / 1000)\n\n    // End call callback\n    this.config.onEnd?.({\n      conversation: this.conversation.slice(1), // Remove system message\n      duration,\n    })\n\n    // Unset params\n    // @ts-ignore\n    this.socket = null\n    // @ts-ignore\n    this.config = null\n  }\n\n  private async onMessage(message: Buffer) {\n    if (!Buffer.isBuffer(message)) {\n      console.warn(`[WS] Message is not a buffer`)\n      return\n    }\n\n    // Commands\n    if (message.byteLength < 15) {\n      const cmd = message.toString()\n      this.debug(`Command: ${cmd}`)\n\n      if (cmd === CallClientCommands.StartSpeaking) {\n        // User started speaking\n        this.isSpeaking = true\n        // Abort answer if there is generation in progress\n        this.abortAnswer = true\n      } else if (cmd === CallClientCommands.Mute) {\n        // User muted the call\n        // Abort answer if there is generation in progress\n        this.abortAnswer = true\n      } else if (cmd === CallClientCommands.StopSpeaking) {\n        // User stopped speaking\n        this.isSpeaking = false\n        await this.onStopSpeaking()\n      }\n    }\n\n    // Audio chunk\n    else if (Buffer.isBuffer(message) && this.isSpeaking) {\n      this.debug(`Received chunk (${message.byteLength} bytes)`)\n      this.chunks.push(message)\n    }\n  }\n\n  private async onStopSpeaking() {\n    // Do nothing if there is no chunk\n    if (this.chunks.length === 0) return\n\n    this.abortAnswer = false\n\n    // Combine audio blob\n    const blob = new Blob(this.chunks, { type: 'audio/ogg' })\n\n    // Reset chunks for next user speech\n    this.chunks.length = 0\n\n    try {\n      // Save file to disk\n      if (debugSaveFile) {\n        const filePath = path.join(__dirname, 'speech.ogg')\n        fs.writeFileSync(filePath, Buffer.from(await blob.arrayBuffer()))\n        return\n      }\n\n      // STT: Get transcript and send to client\n      const transcript = await this.config.speech2Text(\n        blob,\n        this.conversation[this.conversation.length - 1]?.content\n      )\n      if (!transcript) {\n        this.debug('Ignoring empty transcript')\n        return\n      }\n\n      this.debug('User transcript:', transcript)\n\n      // Send transcript to client\n      this.addMessage({ role: 'user', content: transcript })\n\n      if (this.abortAnswer) {\n        this.debug('Answer aborted, no answer generated')\n        return\n      }\n\n      // LLM: Generate answer\n      const answer = await this.config.generateAnswer(this.conversation)\n      if (this.abortAnswer) {\n        this.debug('Answer aborted, ignoring answer')\n        return\n      }\n\n      await this.answer(answer)\n    } catch (error) {\n      console.error('[WS]', error)\n      this.socket.close()\n    }\n  }\n\n  private async answer(message: string) {\n    let isEnd = false\n\n    // Detect end of interview\n    if (message.includes(END_INTERVIEW)) {\n      message = message.replace(END_INTERVIEW, '').trim()\n      isEnd = true\n    }\n\n    if (message.length) {\n      // Send answer to client\n      this.debug('Assistant message:', message)\n      this.addMessage({ role: 'assistant', content: message })\n\n      // TTS: Generate answer audio\n      if (!disableTTS) {\n        const audio = await this.config.text2Speech(message)\n        if (this.abortAnswer) {\n          // Remove last assistant message if aborted\n          this.debug('Answer aborted, removing last assistant message')\n          const lastMessage = this.conversation[this.conversation.length - 1]\n          if (lastMessage?.role === 'assistant') {\n            this.conversation.pop()\n            this.socket.send(CallServerCommands.CancelLastAssistantMessage)\n          }\n          return\n        }\n\n        // Send audio to client\n        this.debug(`Send audio: (${audio.byteLength} bytes)`)\n        this.socket.send(audio)\n      }\n    }\n\n    // End of call\n    if (isEnd) {\n      this.debug('Interview ended')\n      this.socket.send(CallServerCommands.EndInterview)\n    }\n  }\n\n  private debug(...message: any[]) {\n    if (!debugEnabled) return\n    const nowTime = Date.now()\n    console.log(\n      `[WS] [${nowTime - this.startTime} | ${nowTime - this.lastDebug}ms]`,\n      ...message\n    )\n    this.lastDebug = nowTime\n  }\n}\n","export enum CallClientCommands {\n  StartSpeaking = 'startSpeaking',\n  StopSpeaking = 'stopSpeaking',\n  Mute = 'mute',\n}\n\nexport enum CallServerCommands {\n  UserMessage = 'userMessage',\n  AssistantMessage = 'assistantMessage',\n  CancelLastAssistantMessage = 'cancelLastAssistantMessage',\n  EndInterview = 'endInterview',\n}\n\nexport interface CallConfig {\n  systemPrompt: string\n  firstMessage?: string\n  generateAnswer(conversation: Conversation): Promise<string>\n  speech2Text(blob: Blob, prompt?: string): Promise<string>\n  text2Speech(text: string): Promise<ArrayBuffer>\n  onMessage?(message: ConversationMessage): void\n  onEnd?(call: CallSummary): void\n}\n\nexport interface CallSummary {\n  conversation: Conversation\n  duration: number\n}\n\nexport type Conversation = ConversationMessage[]\n\nexport interface ConversationMessage {\n  role: 'system' | 'user' | 'assistant'\n  content: string\n}\n","import WebSocket from 'ws'\n\nexport enum CallErrorCode {\n  BadRequest = 4400,\n  Unauthorized = 4401,\n  NotFound = 4404,\n}\n\nexport class CallError extends Error {\n  code: number\n\n  constructor(code: number, message: string) {\n    super(message)\n    this.code = code\n  }\n}\n\nexport function handleError(socket: WebSocket, error: unknown) {\n  if (error instanceof CallError) {\n    socket.close(error.code, error.message)\n  } else {\n    console.error(error)\n    socket.close(1011)\n  }\n  socket.terminate()\n}\n","import { WebSocket } from 'ws'\nimport { CallError, CallErrorCode } from './errors'\n\nexport async function waitForParams<CallParams>(\n  socket: WebSocket,\n  validate: (params: any) => CallParams\n): Promise<CallParams> {\n  return new Promise<CallParams>((resolve, reject) => {\n    // Handle timeout\n    const timeout = setTimeout(() => {\n      reject(new CallError(CallErrorCode.BadRequest, 'Missing params'))\n    }, 3000)\n\n    const onParams = (payload: string) => {\n      // Clear timeout and listener\n      clearTimeout(timeout)\n      socket.off('message', onParams)\n\n      try {\n        // Parse JSON payload\n        const params = validate(JSON.parse(payload))\n        resolve(params)\n      } catch (error) {\n        reject(new CallError(CallErrorCode.BadRequest, 'Invalid params'))\n      }\n    }\n\n    // Listen for params\n    socket.on('message', onParams)\n  })\n}\n"],"mappings":";AAAA,YAAY,QAAQ;AACpB,YAAY,UAAU;;;ACDf,IAAK,qBAAL,kBAAKA,wBAAL;AACL,EAAAA,oBAAA,mBAAgB;AAChB,EAAAA,oBAAA,kBAAe;AACf,EAAAA,oBAAA,UAAO;AAHG,SAAAA;AAAA,GAAA;AAML,IAAK,qBAAL,kBAAKC,wBAAL;AACL,EAAAA,oBAAA,iBAAc;AACd,EAAAA,oBAAA,sBAAmB;AACnB,EAAAA,oBAAA,gCAA6B;AAC7B,EAAAA,oBAAA,kBAAe;AAJL,SAAAA;AAAA,GAAA;;;ADKZ,IAAM,eAAe;AACrB,IAAM,gBAAgB;AACtB,IAAM,aAAa;AAEZ,IAAM,gBAAgB;AAEtB,IAAM,aAAN,MAAiB;AAAA,EAgBtB,YACS,QACA,QACP;AAFO;AACA;AAjBT,SAAQ,YAAY,KAAK,IAAI;AAC7B,SAAQ,YAAY,KAAK,IAAI;AAG7B;AAAA,SAAQ,cAAc;AAGtB;AAAA,SAAQ,aAAa;AAGrB;AAAA,SAAQ,SAAmB,CAAC;AAS1B,SAAK,eAAe,CAAC,EAAE,MAAM,UAAU,SAAS,OAAO,aAAa,CAAC;AACrE,SAAK,MAAM,cAAc;AAKzB,QAAI,OAAO,cAAc;AACvB,WAAK,OAAO,OAAO,YAAY;AAAA,IACjC,OAAO;AACL,WAAK,OACF,eAAe,KAAK,YAAY,EAChC,KAAK,CAAC,WAAW,KAAK,OAAO,MAAM,CAAC,EACpC,MAAM,CAAC,UAAU;AAChB,gBAAQ,MAAM,QAAQ,KAAK;AAC3B,aAAK,OAAO,MAAM;AAAA,MACpB,CAAC;AAAA,IACL;AAGA,SAAK,OAAO,GAAG,SAAS,KAAK,QAAQ,KAAK,IAAI,CAAC;AAC/C,SAAK,OAAO,GAAG,WAAW,KAAK,UAAU,KAAK,IAAI,CAAC;AAAA,EACrD;AAAA,EAEQ,WAAW,SAA8B;AAC/C,SAAK,aAAa,KAAK,OAAO;AAC9B,SAAK,OAAO;AAAA,MACV,GACE,QAAQ,SAAS,oFAGnB,IAAI,QAAQ,OAAO;AAAA,IACrB;AACA,SAAK,OAAO,YAAY,OAAO;AAAA,EACjC;AAAA,EAEQ,UAAU;AAChB,SAAK,MAAM,mBAAmB;AAC9B,SAAK,cAAc;AACnB,UAAM,WAAW,KAAK,OAAO,KAAK,IAAI,IAAI,KAAK,aAAa,GAAI;AAGhE,SAAK,OAAO,QAAQ;AAAA,MAClB,cAAc,KAAK,aAAa,MAAM,CAAC;AAAA;AAAA,MACvC;AAAA,IACF,CAAC;AAID,SAAK,SAAS;AAEd,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,MAAc,UAAU,SAAiB;AACvC,QAAI,CAAC,OAAO,SAAS,OAAO,GAAG;AAC7B,cAAQ,KAAK,8BAA8B;AAC3C;AAAA,IACF;AAGA,QAAI,QAAQ,aAAa,IAAI;AAC3B,YAAM,MAAM,QAAQ,SAAS;AAC7B,WAAK,MAAM,YAAY,GAAG,EAAE;AAE5B,UAAI,6CAA0C;AAE5C,aAAK,aAAa;AAElB,aAAK,cAAc;AAAA,MACrB,WAAW,2BAAiC;AAG1C,aAAK,cAAc;AAAA,MACrB,WAAW,2CAAyC;AAElD,aAAK,aAAa;AAClB,cAAM,KAAK,eAAe;AAAA,MAC5B;AAAA,IACF,WAGS,OAAO,SAAS,OAAO,KAAK,KAAK,YAAY;AACpD,WAAK,MAAM,mBAAmB,QAAQ,UAAU,SAAS;AACzD,WAAK,OAAO,KAAK,OAAO;AAAA,IAC1B;AAAA,EACF;AAAA,EAEA,MAAc,iBAAiB;AAE7B,QAAI,KAAK,OAAO,WAAW,EAAG;AAE9B,SAAK,cAAc;AAGnB,UAAM,OAAO,IAAI,KAAK,KAAK,QAAQ,EAAE,MAAM,YAAY,CAAC;AAGxD,SAAK,OAAO,SAAS;AAErB,QAAI;AAEF,UAAI,eAAe;AACjB,cAAM,WAAgB,UAAK,WAAW,YAAY;AAClD,QAAG,iBAAc,UAAU,OAAO,KAAK,MAAM,KAAK,YAAY,CAAC,CAAC;AAChE;AAAA,MACF;AAGA,YAAM,aAAa,MAAM,KAAK,OAAO;AAAA,QACnC;AAAA,QACA,KAAK,aAAa,KAAK,aAAa,SAAS,CAAC,GAAG;AAAA,MACnD;AACA,UAAI,CAAC,YAAY;AACf,aAAK,MAAM,2BAA2B;AACtC;AAAA,MACF;AAEA,WAAK,MAAM,oBAAoB,UAAU;AAGzC,WAAK,WAAW,EAAE,MAAM,QAAQ,SAAS,WAAW,CAAC;AAErD,UAAI,KAAK,aAAa;AACpB,aAAK,MAAM,qCAAqC;AAChD;AAAA,MACF;AAGA,YAAM,SAAS,MAAM,KAAK,OAAO,eAAe,KAAK,YAAY;AACjE,UAAI,KAAK,aAAa;AACpB,aAAK,MAAM,iCAAiC;AAC5C;AAAA,MACF;AAEA,YAAM,KAAK,OAAO,MAAM;AAAA,IAC1B,SAAS,OAAO;AACd,cAAQ,MAAM,QAAQ,KAAK;AAC3B,WAAK,OAAO,MAAM;AAAA,IACpB;AAAA,EACF;AAAA,EAEA,MAAc,OAAO,SAAiB;AACpC,QAAI,QAAQ;AAGZ,QAAI,QAAQ,SAAS,aAAa,GAAG;AACnC,gBAAU,QAAQ,QAAQ,eAAe,EAAE,EAAE,KAAK;AAClD,cAAQ;AAAA,IACV;AAEA,QAAI,QAAQ,QAAQ;AAElB,WAAK,MAAM,sBAAsB,OAAO;AACxC,WAAK,WAAW,EAAE,MAAM,aAAa,SAAS,QAAQ,CAAC;AAGvD,UAAI,CAAC,YAAY;AACf,cAAM,QAAQ,MAAM,KAAK,OAAO,YAAY,OAAO;AACnD,YAAI,KAAK,aAAa;AAEpB,eAAK,MAAM,iDAAiD;AAC5D,gBAAM,cAAc,KAAK,aAAa,KAAK,aAAa,SAAS,CAAC;AAClE,cAAI,aAAa,SAAS,aAAa;AACrC,iBAAK,aAAa,IAAI;AACtB,iBAAK,OAAO,kEAAkD;AAAA,UAChE;AACA;AAAA,QACF;AAGA,aAAK,MAAM,gBAAgB,MAAM,UAAU,SAAS;AACpD,aAAK,OAAO,KAAK,KAAK;AAAA,MACxB;AAAA,IACF;AAGA,QAAI,OAAO;AACT,WAAK,MAAM,iBAAiB;AAC5B,WAAK,OAAO,sCAAoC;AAAA,IAClD;AAAA,EACF;AAAA,EAEQ,SAAS,SAAgB;AAC/B,QAAI,CAAC,aAAc;AACnB,UAAM,UAAU,KAAK,IAAI;AACzB,YAAQ;AAAA,MACN,SAAS,UAAU,KAAK,SAAS,MAAM,UAAU,KAAK,SAAS;AAAA,MAC/D,GAAG;AAAA,IACL;AACA,SAAK,YAAY;AAAA,EACnB;AACF;;;AElOO,IAAK,gBAAL,kBAAKC,mBAAL;AACL,EAAAA,8BAAA,gBAAa,QAAb;AACA,EAAAA,8BAAA,kBAAe,QAAf;AACA,EAAAA,8BAAA,cAAW,QAAX;AAHU,SAAAA;AAAA,GAAA;AAML,IAAM,YAAN,cAAwB,MAAM;AAAA,EAGnC,YAAY,MAAc,SAAiB;AACzC,UAAM,OAAO;AACb,SAAK,OAAO;AAAA,EACd;AACF;AAEO,SAAS,YAAY,QAAmB,OAAgB;AAC7D,MAAI,iBAAiB,WAAW;AAC9B,WAAO,MAAM,MAAM,MAAM,MAAM,OAAO;AAAA,EACxC,OAAO;AACL,YAAQ,MAAM,KAAK;AACnB,WAAO,MAAM,IAAI;AAAA,EACnB;AACA,SAAO,UAAU;AACnB;;;ACtBA,eAAsB,cACpB,QACA,UACqB;AACrB,SAAO,IAAI,QAAoB,CAAC,SAAS,WAAW;AAElD,UAAM,UAAU,WAAW,MAAM;AAC/B,aAAO,IAAI,iCAAoC,gBAAgB,CAAC;AAAA,IAClE,GAAG,GAAI;AAEP,UAAM,WAAW,CAAC,YAAoB;AAEpC,mBAAa,OAAO;AACpB,aAAO,IAAI,WAAW,QAAQ;AAE9B,UAAI;AAEF,cAAM,SAAS,SAAS,KAAK,MAAM,OAAO,CAAC;AAC3C,gBAAQ,MAAM;AAAA,MAChB,SAAS,OAAO;AACd,eAAO,IAAI,iCAAoC,gBAAgB,CAAC;AAAA,MAClE;AAAA,IACF;AAGA,WAAO,GAAG,WAAW,QAAQ;AAAA,EAC/B,CAAC;AACH;","names":["CallClientCommands","CallServerCommands","CallErrorCode"]}

package/package.json

@@ -0,0 +1,45 @@
+{
+  "name": "@micdrop/server",
+  "version": "1.0.0",
+  "description": "A lib for Node.js that helps to use the mic and speaker for voice conversation",
+  "author": "Lonestone",
+  "license": "MIT",
+  "main": "./dist/index.js",
+  "module": "./dist/index.mjs",
+  "types": "./dist/index.d.ts",
+  "scripts": {
+    "build": "tsup",
+    "dev": "tsup --watch",
+    "clean": "rm -rf dist",
+    "typecheck": "tsc --noEmit",
+    "prepublishOnly": "npm run build"
+  },
+  "files": [
+    "dist"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/lonestone/micdrop.git",
+    "directory": "packages/server"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "keywords": [
+    "microphone",
+    "audio",
+    "browser",
+    "voice",
+    "conversation"
+  ],
+  "dependencies": {
+    "ws": "^8.18.0"
+  },
+  "devDependencies": {
+    "@types/node": "^20.0.0",
+    "@types/ws": "^8.5.14",
+    "eslint": "^8.0.0",
+    "tsup": "^8.0.0",
+    "typescript": "^5.0.0"
+  }
+}