@mastra/voice-openai-realtime 0.1.0-alpha.1 → 0.1.0-alpha.3

package/.turbo/turbo-build.log

@@ -1,23 +1,23 @@
 
-> @mastra/voice-openai-realtime@0.1.0-alpha.1 build /home/runner/work/mastra/mastra/voice/openai-realtime-api
+> @mastra/voice-openai-realtime@0.1.0-alpha.3 build /home/runner/work/mastra/mastra/voice/openai-realtime-api
 > tsup src/index.ts --format esm,cjs --experimental-dts --clean --treeshake
 
 CLI Building entry: src/index.ts
 CLI Using tsconfig: tsconfig.json
 CLI tsup v8.4.0
 TSC Build start
-TSC ⚡️ Build success in 9409ms
+TSC ⚡️ Build success in 8539ms
 DTS Build start
 CLI Target: es2022
 Analysis will use the bundled TypeScript version 5.8.2
 Writing package typings: /home/runner/work/mastra/mastra/voice/openai-realtime-api/dist/_tsup-dts-rollup.d.ts
 Analysis will use the bundled TypeScript version 5.8.2
 Writing package typings: /home/runner/work/mastra/mastra/voice/openai-realtime-api/dist/_tsup-dts-rollup.d.cts
-DTS ⚡️ Build success in 10803ms
+DTS ⚡️ Build success in 12102ms
 CLI Cleaning output folder
 ESM Build start
 CJS Build start
-CJS dist/index.cjs 17.77 KB
-CJS ⚡️ Build success in 694ms
-ESM dist/index.js 17.72 KB
-ESM ⚡️ Build success in 695ms
+ESM dist/index.js 18.38 KB
+ESM ⚡️ Build success in 718ms
+CJS dist/index.cjs 18.44 KB
+CJS ⚡️ Build success in 718ms

package/CHANGELOG.md

@@ -1,5 +1,24 @@
 # @mastra/voice-openai-realtime
 
+## 0.1.0-alpha.3
+
+### Patch Changes
+
+- a4686e8: Realtime event queue
+- Updated dependencies [b3b34f5]
+- Updated dependencies [a4686e8]
+  - @mastra/core@0.7.0-alpha.3
+
+## 0.1.0-alpha.2
+
+### Patch Changes
+
+- Updated dependencies [a838fde]
+- Updated dependencies [a8bd4cf]
+- Updated dependencies [7a3eeb0]
+- Updated dependencies [6530ad1]
+  - @mastra/core@0.7.0-alpha.2
+
 ## 0.1.0-alpha.1
 
 ### Minor Changes

package/dist/_tsup-dts-rollup.d.cts

@@ -50,14 +50,15 @@ export declare class OpenAIRealtimeVoice extends MastraVoice {
     private instructions?;
     private tools?;
     private debug;
+    private queue;
+    private transcriber;
     /**
      * Creates a new instance of OpenAIRealtimeVoice.
      *
      * @param options - Configuration options for the voice instance
-     * @param options.chatModel - Configuration for the chat model
-     * @param options.chatModel.model - The model ID to use (defaults to GPT-4 Mini Realtime)
-     * @param options.chatModel.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
-     * @param options.chatModel.tools - Tools configuration for the model
+     * @param options.url - The base URL for the OpenAI Realtime API
+     * @param options.model - The model ID to use (defaults to GPT-4 Mini Realtime)
+     * @param options.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
      * @param options.speaker - Voice ID to use (defaults to 'alloy')
      * @param options.debug - Enable debug mode
      *
@@ -72,15 +73,12 @@ export declare class OpenAIRealtimeVoice extends MastraVoice {
      * });
      * ```
      */
-    constructor({ chatModel, speaker, debug, }?: {
-        chatModel?: {
-            model?: string;
-            apiKey?: string;
-            tools?: TTools;
-            instructions?: string;
-            url?: string;
-        };
+    constructor(options?: {
+        model?: string;
+        url?: string;
+        apiKey?: string;
         speaker?: Realtime.Voice;
+        transcriber?: Realtime.AudioTranscriptionModel;
         debug?: boolean;
     });
     /**
@@ -108,6 +106,19 @@ export declare class OpenAIRealtimeVoice extends MastraVoice {
      * ```
      */
     close(): void;
+    /**
+     * Equips the voice instance with a set of instructions.
+     * Instructions allow the model to perform additional actions during conversations.
+     *
+     * @param instructions - Optional instructions to addInstructions
+     * @returns Transformed instructions ready for use with the model
+     *
+     * @example
+     * ```typescript
+     * voice.addInstuctions('You are a helpful assistant.');
+     * ```
+     */
+    addInstructions(instructions?: string): void;
     /**
      * Equips the voice instance with a set of tools.
      * Tools allow the model to perform additional actions during conversations.

package/dist/_tsup-dts-rollup.d.ts

@@ -50,14 +50,15 @@ export declare class OpenAIRealtimeVoice extends MastraVoice {
     private instructions?;
     private tools?;
     private debug;
+    private queue;
+    private transcriber;
     /**
      * Creates a new instance of OpenAIRealtimeVoice.
      *
      * @param options - Configuration options for the voice instance
-     * @param options.chatModel - Configuration for the chat model
-     * @param options.chatModel.model - The model ID to use (defaults to GPT-4 Mini Realtime)
-     * @param options.chatModel.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
-     * @param options.chatModel.tools - Tools configuration for the model
+     * @param options.url - The base URL for the OpenAI Realtime API
+     * @param options.model - The model ID to use (defaults to GPT-4 Mini Realtime)
+     * @param options.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
      * @param options.speaker - Voice ID to use (defaults to 'alloy')
      * @param options.debug - Enable debug mode
      *
@@ -72,15 +73,12 @@ export declare class OpenAIRealtimeVoice extends MastraVoice {
      * });
      * ```
      */
-    constructor({ chatModel, speaker, debug, }?: {
-        chatModel?: {
-            model?: string;
-            apiKey?: string;
-            tools?: TTools;
-            instructions?: string;
-            url?: string;
-        };
+    constructor(options?: {
+        model?: string;
+        url?: string;
+        apiKey?: string;
         speaker?: Realtime.Voice;
+        transcriber?: Realtime.AudioTranscriptionModel;
         debug?: boolean;
     });
     /**
@@ -108,6 +106,19 @@ export declare class OpenAIRealtimeVoice extends MastraVoice {
      * ```
      */
     close(): void;
+    /**
+     * Equips the voice instance with a set of instructions.
+     * Instructions allow the model to perform additional actions during conversations.
+     *
+     * @param instructions - Optional instructions to addInstructions
+     * @returns Transformed instructions ready for use with the model
+     *
+     * @example
+     * ```typescript
+     * voice.addInstuctions('You are a helpful assistant.');
+     * ```
+     */
+    addInstructions(instructions?: string): void;
     /**
      * Equips the voice instance with a set of tools.
      * Tools allow the model to perform additional actions during conversations.

package/dist/index.cjs

@@ -1,10 +1,10 @@
 'use strict';
 
-var voice = require('@mastra/core/voice');
+var events = require('events');
 var stream = require('stream');
-var zodToJsonSchema = require('zod-to-json-schema');
+var voice = require('@mastra/core/voice');
 var ws = require('ws');
-var events = require('events');
+var zodToJsonSchema = require('zod-to-json-schema');
 
 // src/index.ts
 var transformTools = (tools) => {
@@ -65,7 +65,10 @@ var transformTools = (tools) => {
 var isReadableStream = (obj) => {
   return obj && obj instanceof stream.Readable && typeof obj.read === "function" && typeof obj.pipe === "function" && obj.readable === true;
 };
+
+// src/index.ts
 var DEFAULT_VOICE = "alloy";
+var DEFAULT_TRANSCRIBER = "whisper-1";
 var DEFAULT_URL = "wss://api.openai.com/v1/realtime";
 var DEFAULT_MODEL = "gpt-4o-mini-realtime-preview-2024-12-17";
 var VOICES = ["alloy", "ash", "ballad", "coral", "echo", "sage", "shimmer", "verse"];
@@ -77,14 +80,15 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
   instructions;
   tools;
   debug;
+  queue = [];
+  transcriber;
   /**
    * Creates a new instance of OpenAIRealtimeVoice.
    *
    * @param options - Configuration options for the voice instance
-   * @param options.chatModel - Configuration for the chat model
-   * @param options.chatModel.model - The model ID to use (defaults to GPT-4 Mini Realtime)
-   * @param options.chatModel.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
-   * @param options.chatModel.tools - Tools configuration for the model
+   * @param options.url - The base URL for the OpenAI Realtime API
+   * @param options.model - The model ID to use (defaults to GPT-4 Mini Realtime)
+   * @param options.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
    * @param options.speaker - Voice ID to use (defaults to 'alloy')
    * @param options.debug - Enable debug mode
    *
@@ -99,14 +103,10 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
    * });
    * ```
    */
-  constructor({
-    chatModel,
-    speaker,
-    debug = false
-  } = {}) {
+  constructor(options = {}) {
     super();
-    const url = `${chatModel?.url || DEFAULT_URL}?model=${chatModel?.model || DEFAULT_MODEL}`;
-    const apiKey = chatModel?.apiKey || process.env.OPENAI_API_KEY;
+    const url = `${options.url || DEFAULT_URL}?model=${options.model || DEFAULT_MODEL}`;
+    const apiKey = options.apiKey || process.env.OPENAI_API_KEY;
     this.ws = new ws.WebSocket(url, void 0, {
       headers: {
         Authorization: "Bearer " + apiKey,
@@ -116,10 +116,9 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
     this.client = new events.EventEmitter();
     this.state = "close";
     this.events = {};
-    this.tools = chatModel?.tools;
-    this.instructions = chatModel?.instructions;
-    this.speaker = speaker || DEFAULT_VOICE;
-    this.debug = debug;
+    this.speaker = options.speaker || DEFAULT_VOICE;
+    this.transcriber = options.transcriber || DEFAULT_TRANSCRIBER;
+    this.debug = options.debug || false;
     this.setupEventListeners();
   }
   /**
@@ -150,6 +149,21 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
     this.ws.close();
     this.state = "close";
   }
+  /**
+   * Equips the voice instance with a set of instructions.
+   * Instructions allow the model to perform additional actions during conversations.
+   *
+   * @param instructions - Optional instructions to addInstructions
+   * @returns Transformed instructions ready for use with the model
+   *
+   * @example
+   * ```typescript
+   * voice.addInstuctions('You are a helpful assistant.');
+   * ```
+   */
+  addInstructions(instructions) {
+    this.instructions = instructions;
+  }
   /**
    * Equips the voice instance with a set of tools.
    * Tools allow the model to perform additional actions during conversations.
@@ -167,10 +181,7 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
    * ```
    */
   addTools(tools) {
-    const openaiTools = transformTools(tools);
-    this.updateConfig({
-      tools: openaiTools.map((t) => t.openaiTool)
-    });
+    this.tools = tools || {};
   }
   /**
    * Emits a speaking event using the configured voice model.
@@ -314,7 +325,7 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
       instructions: this.instructions,
       tools: openaiTools.map((t) => t.openaiTool),
       input_audio_transcription: {
-        model: "whisper-1"
+        model: this.transcriber
       },
       voice: this.speaker
     });
@@ -462,6 +473,10 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
     });
     this.client.on("session.created", (ev) => {
       this.emit("session.created", ev);
+      const queue = this.queue.splice(0, this.queue.length);
+      for (const ev2 of queue) {
+        this.ws.send(JSON.stringify(ev2));
+      }
     });
     this.client.on("session.updated", (ev) => {
       this.emit("session.updated", ev);
@@ -474,10 +489,10 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
       this.emit("speaker", speakerStream);
     });
     this.client.on("conversation.item.input_audio_transcription.delta", (ev) => {
-      this.emit("transcribing", { text: ev.delta, response_id: ev.response_id, role: "user" });
+      this.emit("writing", { text: ev.delta, response_id: ev.response_id, role: "user" });
     });
     this.client.on("conversation.item.input_audio_transcription.done", (ev) => {
-      this.emit("transcribing", { text: "\n", response_id: ev.response_id, role: "user" });
+      this.emit("writing", { text: "\n", response_id: ev.response_id, role: "user" });
     });
     this.client.on("response.audio.delta", (ev) => {
       const audio = Buffer.from(ev.delta, "base64");
@@ -491,19 +506,19 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
       stream?.end();
     });
     this.client.on("response.audio_transcript.delta", (ev) => {
-      this.emit("writing", { text: ev.delta, response_id: ev.response_id });
+      this.emit("writing", { text: ev.delta, response_id: ev.response_id, role: "assistant" });
     });
     this.client.on("response.audio_transcript.done", (ev) => {
-      this.emit("writing", { text: "\n", response_id: ev.response_id });
+      this.emit("writing", { text: "\n", response_id: ev.response_id, role: "assistant" });
     });
     this.client.on("response.text.delta", (ev) => {
-      this.emit("writing", { text: ev.delta, response_id: ev.response_id });
+      this.emit("writing", { text: ev.delta, response_id: ev.response_id, role: "assistant" });
     });
     this.client.on("response.text.done", (ev) => {
-      this.emit("writing", { text: "\n", response_id: ev.response_id });
+      this.emit("writing", { text: "\n", response_id: ev.response_id, role: "assistant" });
     });
-    this.client.on("response.done", (ev) => {
-      this.handleFunctionCalls(ev);
+    this.client.on("response.done", async (ev) => {
+      await this.handleFunctionCalls(ev);
       this.emit("response.done", ev);
       speakerStreams.delete(ev.response.id);
     });
@@ -565,12 +580,16 @@ var OpenAIRealtimeVoice = class extends voice.MastraVoice {
     return btoa(binary);
   }
   sendEvent(type, data) {
-    this.ws.send(
-      JSON.stringify({
-        type,
-        ...data
-      })
-    );
+    if (this.ws.readyState !== this.ws.OPEN) {
+      this.queue.push({ type, ...data });
+    } else {
+      this.ws.send(
+        JSON.stringify({
+          type,
+          ...data
+        })
+      );
+    }
   }
 };
 

package/dist/index.js

@@ -1,8 +1,8 @@
-import { MastraVoice } from '@mastra/core/voice';
+import { EventEmitter } from 'events';
 import { PassThrough, Readable } from 'stream';
-import { zodToJsonSchema } from 'zod-to-json-schema';
+import { MastraVoice } from '@mastra/core/voice';
 import { WebSocket } from 'ws';
-import { EventEmitter } from 'events';
+import { zodToJsonSchema } from 'zod-to-json-schema';
 
 // src/index.ts
 var transformTools = (tools) => {
@@ -63,7 +63,10 @@ var transformTools = (tools) => {
 var isReadableStream = (obj) => {
   return obj && obj instanceof Readable && typeof obj.read === "function" && typeof obj.pipe === "function" && obj.readable === true;
 };
+
+// src/index.ts
 var DEFAULT_VOICE = "alloy";
+var DEFAULT_TRANSCRIBER = "whisper-1";
 var DEFAULT_URL = "wss://api.openai.com/v1/realtime";
 var DEFAULT_MODEL = "gpt-4o-mini-realtime-preview-2024-12-17";
 var VOICES = ["alloy", "ash", "ballad", "coral", "echo", "sage", "shimmer", "verse"];
@@ -75,14 +78,15 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
   instructions;
   tools;
   debug;
+  queue = [];
+  transcriber;
   /**
    * Creates a new instance of OpenAIRealtimeVoice.
    *
    * @param options - Configuration options for the voice instance
-   * @param options.chatModel - Configuration for the chat model
-   * @param options.chatModel.model - The model ID to use (defaults to GPT-4 Mini Realtime)
-   * @param options.chatModel.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
-   * @param options.chatModel.tools - Tools configuration for the model
+   * @param options.url - The base URL for the OpenAI Realtime API
+   * @param options.model - The model ID to use (defaults to GPT-4 Mini Realtime)
+   * @param options.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
    * @param options.speaker - Voice ID to use (defaults to 'alloy')
    * @param options.debug - Enable debug mode
    *
@@ -97,14 +101,10 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
    * });
    * ```
    */
-  constructor({
-    chatModel,
-    speaker,
-    debug = false
-  } = {}) {
+  constructor(options = {}) {
     super();
-    const url = `${chatModel?.url || DEFAULT_URL}?model=${chatModel?.model || DEFAULT_MODEL}`;
-    const apiKey = chatModel?.apiKey || process.env.OPENAI_API_KEY;
+    const url = `${options.url || DEFAULT_URL}?model=${options.model || DEFAULT_MODEL}`;
+    const apiKey = options.apiKey || process.env.OPENAI_API_KEY;
     this.ws = new WebSocket(url, void 0, {
       headers: {
         Authorization: "Bearer " + apiKey,
@@ -114,10 +114,9 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
     this.client = new EventEmitter();
     this.state = "close";
     this.events = {};
-    this.tools = chatModel?.tools;
-    this.instructions = chatModel?.instructions;
-    this.speaker = speaker || DEFAULT_VOICE;
-    this.debug = debug;
+    this.speaker = options.speaker || DEFAULT_VOICE;
+    this.transcriber = options.transcriber || DEFAULT_TRANSCRIBER;
+    this.debug = options.debug || false;
     this.setupEventListeners();
   }
   /**
@@ -148,6 +147,21 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
     this.ws.close();
     this.state = "close";
   }
+  /**
+   * Equips the voice instance with a set of instructions.
+   * Instructions allow the model to perform additional actions during conversations.
+   *
+   * @param instructions - Optional instructions to addInstructions
+   * @returns Transformed instructions ready for use with the model
+   *
+   * @example
+   * ```typescript
+   * voice.addInstuctions('You are a helpful assistant.');
+   * ```
+   */
+  addInstructions(instructions) {
+    this.instructions = instructions;
+  }
   /**
    * Equips the voice instance with a set of tools.
    * Tools allow the model to perform additional actions during conversations.
@@ -165,10 +179,7 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
    * ```
    */
   addTools(tools) {
-    const openaiTools = transformTools(tools);
-    this.updateConfig({
-      tools: openaiTools.map((t) => t.openaiTool)
-    });
+    this.tools = tools || {};
   }
   /**
    * Emits a speaking event using the configured voice model.
@@ -312,7 +323,7 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
       instructions: this.instructions,
       tools: openaiTools.map((t) => t.openaiTool),
       input_audio_transcription: {
-        model: "whisper-1"
+        model: this.transcriber
       },
       voice: this.speaker
     });
@@ -460,6 +471,10 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
     });
     this.client.on("session.created", (ev) => {
       this.emit("session.created", ev);
+      const queue = this.queue.splice(0, this.queue.length);
+      for (const ev2 of queue) {
+        this.ws.send(JSON.stringify(ev2));
+      }
     });
     this.client.on("session.updated", (ev) => {
       this.emit("session.updated", ev);
@@ -472,10 +487,10 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
       this.emit("speaker", speakerStream);
     });
     this.client.on("conversation.item.input_audio_transcription.delta", (ev) => {
-      this.emit("transcribing", { text: ev.delta, response_id: ev.response_id, role: "user" });
+      this.emit("writing", { text: ev.delta, response_id: ev.response_id, role: "user" });
     });
     this.client.on("conversation.item.input_audio_transcription.done", (ev) => {
-      this.emit("transcribing", { text: "\n", response_id: ev.response_id, role: "user" });
+      this.emit("writing", { text: "\n", response_id: ev.response_id, role: "user" });
     });
     this.client.on("response.audio.delta", (ev) => {
       const audio = Buffer.from(ev.delta, "base64");
@@ -489,19 +504,19 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
       stream?.end();
     });
     this.client.on("response.audio_transcript.delta", (ev) => {
-      this.emit("writing", { text: ev.delta, response_id: ev.response_id });
+      this.emit("writing", { text: ev.delta, response_id: ev.response_id, role: "assistant" });
     });
     this.client.on("response.audio_transcript.done", (ev) => {
-      this.emit("writing", { text: "\n", response_id: ev.response_id });
+      this.emit("writing", { text: "\n", response_id: ev.response_id, role: "assistant" });
     });
     this.client.on("response.text.delta", (ev) => {
-      this.emit("writing", { text: ev.delta, response_id: ev.response_id });
+      this.emit("writing", { text: ev.delta, response_id: ev.response_id, role: "assistant" });
     });
     this.client.on("response.text.done", (ev) => {
-      this.emit("writing", { text: "\n", response_id: ev.response_id });
+      this.emit("writing", { text: "\n", response_id: ev.response_id, role: "assistant" });
     });
-    this.client.on("response.done", (ev) => {
-      this.handleFunctionCalls(ev);
+    this.client.on("response.done", async (ev) => {
+      await this.handleFunctionCalls(ev);
       this.emit("response.done", ev);
       speakerStreams.delete(ev.response.id);
     });
@@ -563,12 +578,16 @@ var OpenAIRealtimeVoice = class extends MastraVoice {
     return btoa(binary);
   }
   sendEvent(type, data) {
-    this.ws.send(
-      JSON.stringify({
-        type,
-        ...data
-      })
-    );
+    if (this.ws.readyState !== this.ws.OPEN) {
+      this.queue.push({ type, ...data });
+    } else {
+      this.ws.send(
+        JSON.stringify({
+          type,
+          ...data
+        })
+      );
+    }
   }
 };
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mastra/voice-openai-realtime",
-  "version": "0.1.0-alpha.1",
+  "version": "0.1.0-alpha.3",
   "description": "Mastra OpenAI Realtime API integration",
   "type": "module",
   "main": "dist/index.js",
@@ -22,7 +22,7 @@
     "openai-realtime-api": "^1.0.7",
     "ws": "^8.18.1",
     "zod-to-json-schema": "^3.24.1",
-    "@mastra/core": "^0.7.0-alpha.1"
+    "@mastra/core": "^0.7.0-alpha.3"
   },
   "devDependencies": {
     "@microsoft/api-extractor": "^7.49.2",

package/src/index.ts

@@ -1,10 +1,10 @@
+import { EventEmitter } from 'events';
+import { PassThrough } from 'stream';
 import type { ToolsInput } from '@mastra/core/agent';
 import { MastraVoice } from '@mastra/core/voice';
-import { isReadableStream, transformTools } from './utils';
-import { WebSocket } from 'ws';
-import { EventEmitter } from 'events';
 import type { Realtime, RealtimeServerEvents } from 'openai-realtime-api';
-import { PassThrough } from 'stream';
+import { WebSocket } from 'ws';
+import { isReadableStream, transformTools } from './utils';
 
 /**
  * Event callback function type
@@ -29,6 +29,8 @@ type EventMap = {
 /** Default voice for text-to-speech. 'alloy' provides a neutral, balanced voice suitable for most use cases */
 const DEFAULT_VOICE: Realtime.Voice = 'alloy';
 
+const DEFAULT_TRANSCRIBER: Realtime.AudioTranscriptionModel = 'whisper-1';
+
 const DEFAULT_URL = 'wss://api.openai.com/v1/realtime';
 
 /**
@@ -36,21 +38,22 @@ const DEFAULT_URL = 'wss://api.openai.com/v1/realtime';
  * This model is optimized for low-latency responses while maintaining high quality output.
  */
 const DEFAULT_MODEL = 'gpt-4o-mini-realtime-preview-2024-12-17';
-/**
- * Default Voice Activity Detection (VAD) configuration.
- * These settings control how the system detects speech segments.
- *
- * @property {string} type - Uses server-side VAD for better accuracy
- * @property {number} threshold - Speech detection sensitivity (0.5 = balanced)
- * @property {number} prefix_padding_ms - Includes 1 second of audio before speech
- * @property {number} silence_duration_ms - Waits 1 second of silence before ending turn
- */
-const DEFAULT_VAD_CONFIG = {
-  type: 'server_vad',
-  threshold: 0.5,
-  prefix_padding_ms: 1000,
-  silence_duration_ms: 1000,
-} as Realtime.TurnDetection;
+
+// /**
+//  * Default Voice Activity Detection (VAD) configuration.
+//  * These settings control how the system detects speech segments.
+//  *
+//  * @property {string} type - Uses server-side VAD for better accuracy
+//  * @property {number} threshold - Speech detection sensitivity (0.5 = balanced)
+//  * @property {number} prefix_padding_ms - Includes 1 second of audio before speech
+//  * @property {number} silence_duration_ms - Waits 1 second of silence before ending turn
+//  */
+// const DEFAULT_VAD_CONFIG = {
+//   type: 'server_vad',
+//   threshold: 0.5,
+//   prefix_padding_ms: 1000,
+//   silence_duration_ms: 1000,
+// } as Realtime.TurnDetection;
 
 type TTools = ToolsInput;
 
@@ -110,15 +113,16 @@ export class OpenAIRealtimeVoice extends MastraVoice {
   private instructions?: string;
   private tools?: TTools;
   private debug: boolean;
+  private queue: unknown[] = [];
+  private transcriber: Realtime.AudioTranscriptionModel;
 
   /**
    * Creates a new instance of OpenAIRealtimeVoice.
    *
    * @param options - Configuration options for the voice instance
-   * @param options.chatModel - Configuration for the chat model
-   * @param options.chatModel.model - The model ID to use (defaults to GPT-4 Mini Realtime)
-   * @param options.chatModel.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
-   * @param options.chatModel.tools - Tools configuration for the model
+   * @param options.url - The base URL for the OpenAI Realtime API
+   * @param options.model - The model ID to use (defaults to GPT-4 Mini Realtime)
+   * @param options.apiKey - OpenAI API key. Falls back to process.env.OPENAI_API_KEY
    * @param options.speaker - Voice ID to use (defaults to 'alloy')
    * @param options.debug - Enable debug mode
    *
@@ -133,25 +137,20 @@ export class OpenAIRealtimeVoice extends MastraVoice {
    * });
    * ```
    */
-  constructor({
-    chatModel,
-    speaker,
-    debug = false,
-  }: {
-    chatModel?: {
+  constructor(
+    options: {
       model?: string;
-      apiKey?: string;
-      tools?: TTools;
-      instructions?: string;
       url?: string;
-    };
-    speaker?: Realtime.Voice;
-    debug?: boolean;
-  } = {}) {
+      apiKey?: string;
+      speaker?: Realtime.Voice;
+      transcriber?: Realtime.AudioTranscriptionModel;
+      debug?: boolean;
+    } = {},
+  ) {
     super();
 
-    const url = `${chatModel?.url || DEFAULT_URL}?model=${chatModel?.model || DEFAULT_MODEL}`;
-    const apiKey = chatModel?.apiKey || process.env.OPENAI_API_KEY;
+    const url = `${options.url || DEFAULT_URL}?model=${options.model || DEFAULT_MODEL}`;
+    const apiKey = options.apiKey || process.env.OPENAI_API_KEY;
     this.ws = new WebSocket(url, undefined, {
       headers: {
         Authorization: 'Bearer ' + apiKey,
@@ -162,10 +161,9 @@ export class OpenAIRealtimeVoice extends MastraVoice {
     this.client = new EventEmitter();
     this.state = 'close';
     this.events = {} as EventMap;
-    this.tools = chatModel?.tools;
-    this.instructions = chatModel?.instructions;
-    this.speaker = speaker || DEFAULT_VOICE;
-    this.debug = debug;
+    this.speaker = options.speaker || DEFAULT_VOICE;
+    this.transcriber = options.transcriber || DEFAULT_TRANSCRIBER;
+    this.debug = options.debug || false;
     this.setupEventListeners();
   }
 
@@ -199,6 +197,22 @@ export class OpenAIRealtimeVoice extends MastraVoice {
     this.state = 'close';
   }
 
+  /**
+   * Equips the voice instance with a set of instructions.
+   * Instructions allow the model to perform additional actions during conversations.
+   *
+   * @param instructions - Optional instructions to addInstructions
+   * @returns Transformed instructions ready for use with the model
+   *
+   * @example
+   * ```typescript
+   * voice.addInstuctions('You are a helpful assistant.');
+   * ```
+   */
+  addInstructions(instructions?: string) {
+    this.instructions = instructions;
+  }
+
   /**
    * Equips the voice instance with a set of tools.
    * Tools allow the model to perform additional actions during conversations.
@@ -216,10 +230,7 @@ export class OpenAIRealtimeVoice extends MastraVoice {
    * ```
    */
   addTools(tools?: TTools) {
-    const openaiTools = transformTools(tools);
-    this.updateConfig({
-      tools: openaiTools.map(t => t.openaiTool),
-    });
+    this.tools = tools || {};
   }
 
   /**
@@ -375,7 +386,7 @@ export class OpenAIRealtimeVoice extends MastraVoice {
       instructions: this.instructions,
       tools: openaiTools.map(t => t.openaiTool),
       input_audio_transcription: {
-        model: 'whisper-1',
+        model: this.transcriber,
       },
       voice: this.speaker,
     });
@@ -535,6 +546,11 @@ export class OpenAIRealtimeVoice extends MastraVoice {
 
     this.client.on('session.created', ev => {
       this.emit('session.created', ev);
+
+      const queue = this.queue.splice(0, this.queue.length);
+      for (const ev of queue) {
+        this.ws.send(JSON.stringify(ev));
+      }
     });
     this.client.on('session.updated', ev => {
       this.emit('session.updated', ev);
@@ -550,10 +566,10 @@ export class OpenAIRealtimeVoice extends MastraVoice {
       this.emit('speaker', speakerStream);
     });
     this.client.on('conversation.item.input_audio_transcription.delta', ev => {
-      this.emit('transcribing', { text: ev.delta, response_id: ev.response_id, role: 'user' });
+      this.emit('writing', { text: ev.delta, response_id: ev.response_id, role: 'user' });
     });
     this.client.on('conversation.item.input_audio_transcription.done', ev => {
-      this.emit('transcribing', { text: '\n', response_id: ev.response_id, role: 'user' });
+      this.emit('writing', { text: '\n', response_id: ev.response_id, role: 'user' });
     });
     this.client.on('response.audio.delta', ev => {
       const audio = Buffer.from(ev.delta, 'base64');
@@ -569,19 +585,19 @@ export class OpenAIRealtimeVoice extends MastraVoice {
       stream?.end();
     });
     this.client.on('response.audio_transcript.delta', ev => {
-      this.emit('writing', { text: ev.delta, response_id: ev.response_id });
+      this.emit('writing', { text: ev.delta, response_id: ev.response_id, role: 'assistant' });
     });
     this.client.on('response.audio_transcript.done', ev => {
-      this.emit('writing', { text: '\n', response_id: ev.response_id });
+      this.emit('writing', { text: '\n', response_id: ev.response_id, role: 'assistant' });
     });
     this.client.on('response.text.delta', ev => {
-      this.emit('writing', { text: ev.delta, response_id: ev.response_id });
+      this.emit('writing', { text: ev.delta, response_id: ev.response_id, role: 'assistant' });
     });
     this.client.on('response.text.done', ev => {
-      this.emit('writing', { text: '\n', response_id: ev.response_id });
+      this.emit('writing', { text: '\n', response_id: ev.response_id, role: 'assistant' });
     });
-    this.client.on('response.done', ev => {
-      this.handleFunctionCalls(ev);
+    this.client.on('response.done', async ev => {
+      await this.handleFunctionCalls(ev);
       this.emit('response.done', ev);
       speakerStreams.delete(ev.response.id);
     });
@@ -647,11 +663,15 @@ export class OpenAIRealtimeVoice extends MastraVoice {
   }
 
   private sendEvent(type: string, data: any) {
-    this.ws.send(
-      JSON.stringify({
-        type: type,
-        ...data,
-      }),
-    );
+    if (this.ws.readyState !== this.ws.OPEN) {
+      this.queue.push({ type: type, ...data });
+    } else {
+      this.ws.send(
+        JSON.stringify({
+          type: type,
+          ...data,
+        }),
+      );
+    }
   }
 }