@lokutor/sdk 1.1.11 → 1.1.12

package/dist/index.d.mts

@@ -97,6 +97,28 @@ interface Viseme {
     c: string;
     t: number;
 }
+/**
+ * Tool definition for LLM function calling (OpenAI format)
+ */
+interface ToolDefinition {
+    type: 'function';
+    function: {
+        name: string;
+        description: string;
+        parameters: {
+            type: 'object';
+            properties: Record<string, any>;
+            required?: string[];
+        };
+    };
+}
+/**
+ * Event data for tool execution
+ */
+interface ToolCall {
+    name: string;
+    arguments: string;
+}
 
 /**
  * Main client for Lokutor Voice Agent SDK
@@ -109,6 +131,7 @@ declare class VoiceAgentClient {
     prompt: string;
     voice: VoiceStyle;
     language: Language;
+    tools: ToolDefinition[];
     private onTranscription?;
     private onResponse?;
     private onAudioCallback?;
@@ -121,6 +144,7 @@ declare class VoiceAgentClient {
     private wantVisemes;
     private audioManager;
     private enableAudio;
+    private currentGeneration;
     private isUserDisconnect;
     private reconnecting;
     private reconnectAttempts;
@@ -132,6 +156,7 @@ declare class VoiceAgentClient {
         visemes?: boolean;
         onVisemes?: (visemes: Viseme[]) => void;
         enableAudio?: boolean;
+        tools?: ToolDefinition[];
     });
     /**
      * Connect to the Lokutor Voice Agent server
@@ -418,4 +443,4 @@ declare class BrowserAudioManager {
     isRecording(): boolean;
 }
 
-export { AUDIO_CONFIG, type AnalyserConfig, type BrowserAudioConfig, BrowserAudioManager, type BrowserAudioOptions, DEFAULT_URLS, Language, type LokutorConfig, StreamResampler, type SynthesizeOptions, TTSClient, type Viseme, VoiceAgentClient, type VoiceAgentOptions, VoiceStyle, applyLowPassFilter, bytesToPcm16, calculateRMS, float32ToPcm16, normalizeAudio, pcm16ToBytes, pcm16ToFloat32, resample, resampleWithAntiAliasing, simpleConversation, simpleTTS };
+export { AUDIO_CONFIG, type AnalyserConfig, type BrowserAudioConfig, BrowserAudioManager, type BrowserAudioOptions, DEFAULT_URLS, Language, type LokutorConfig, StreamResampler, type SynthesizeOptions, TTSClient, type ToolCall, type ToolDefinition, type Viseme, VoiceAgentClient, type VoiceAgentOptions, VoiceStyle, applyLowPassFilter, bytesToPcm16, calculateRMS, float32ToPcm16, normalizeAudio, pcm16ToBytes, pcm16ToFloat32, resample, resampleWithAntiAliasing, simpleConversation, simpleTTS };

package/dist/index.d.ts

@@ -97,6 +97,28 @@ interface Viseme {
     c: string;
     t: number;
 }
+/**
+ * Tool definition for LLM function calling (OpenAI format)
+ */
+interface ToolDefinition {
+    type: 'function';
+    function: {
+        name: string;
+        description: string;
+        parameters: {
+            type: 'object';
+            properties: Record<string, any>;
+            required?: string[];
+        };
+    };
+}
+/**
+ * Event data for tool execution
+ */
+interface ToolCall {
+    name: string;
+    arguments: string;
+}
 
 /**
  * Main client for Lokutor Voice Agent SDK
@@ -109,6 +131,7 @@ declare class VoiceAgentClient {
     prompt: string;
     voice: VoiceStyle;
     language: Language;
+    tools: ToolDefinition[];
     private onTranscription?;
     private onResponse?;
     private onAudioCallback?;
@@ -121,6 +144,7 @@ declare class VoiceAgentClient {
     private wantVisemes;
     private audioManager;
     private enableAudio;
+    private currentGeneration;
     private isUserDisconnect;
     private reconnecting;
     private reconnectAttempts;
@@ -132,6 +156,7 @@ declare class VoiceAgentClient {
         visemes?: boolean;
         onVisemes?: (visemes: Viseme[]) => void;
         enableAudio?: boolean;
+        tools?: ToolDefinition[];
     });
     /**
      * Connect to the Lokutor Voice Agent server
@@ -418,4 +443,4 @@ declare class BrowserAudioManager {
     isRecording(): boolean;
 }
 
-export { AUDIO_CONFIG, type AnalyserConfig, type BrowserAudioConfig, BrowserAudioManager, type BrowserAudioOptions, DEFAULT_URLS, Language, type LokutorConfig, StreamResampler, type SynthesizeOptions, TTSClient, type Viseme, VoiceAgentClient, type VoiceAgentOptions, VoiceStyle, applyLowPassFilter, bytesToPcm16, calculateRMS, float32ToPcm16, normalizeAudio, pcm16ToBytes, pcm16ToFloat32, resample, resampleWithAntiAliasing, simpleConversation, simpleTTS };
+export { AUDIO_CONFIG, type AnalyserConfig, type BrowserAudioConfig, BrowserAudioManager, type BrowserAudioOptions, DEFAULT_URLS, Language, type LokutorConfig, StreamResampler, type SynthesizeOptions, TTSClient, type ToolCall, type ToolDefinition, type Viseme, VoiceAgentClient, type VoiceAgentOptions, VoiceStyle, applyLowPassFilter, bytesToPcm16, calculateRMS, float32ToPcm16, normalizeAudio, pcm16ToBytes, pcm16ToFloat32, resample, resampleWithAntiAliasing, simpleConversation, simpleTTS };

package/dist/index.js

@@ -506,6 +506,7 @@ var VoiceAgentClient = class {
   prompt;
   voice;
   language;
+  tools = [];
   // Callbacks
   onTranscription;
   onResponse;
@@ -519,6 +520,7 @@ var VoiceAgentClient = class {
   wantVisemes = false;
   audioManager = null;
   enableAudio = false;
+  currentGeneration = 0;
   // Connection resilience
   isUserDisconnect = false;
   reconnecting = false;
@@ -537,6 +539,7 @@ var VoiceAgentClient = class {
     this.onError = config.onError;
     this.wantVisemes = config.visemes || false;
     this.enableAudio = config.enableAudio ?? false;
+    this.tools = config.tools || [];
   }
   /**
    * Connect to the Lokutor Voice Agent server
@@ -617,7 +620,10 @@ var VoiceAgentClient = class {
     this.ws.send(JSON.stringify({ type: "voice", data: this.voice }));
     this.ws.send(JSON.stringify({ type: "language", data: this.language }));
     this.ws.send(JSON.stringify({ type: "visemes", data: this.wantVisemes }));
-    console.log(`\u2699\uFE0F Configured: voice=${this.voice}, language=${this.language}, visemes=${this.wantVisemes}`);
+    if (this.tools && this.tools.length > 0) {
+      this.ws.send(JSON.stringify({ type: "tools", data: this.tools }));
+    }
+    console.log(`\u2699\uFE0F Configured: voice=${this.voice}, language=${this.language}, visemes=${this.wantVisemes}, tools=${this.tools.length}`);
   }
   /**
    * Send raw PCM audio data to the server
@@ -631,7 +637,11 @@ var VoiceAgentClient = class {
   /**
    * Handle incoming binary data (audio response)
    */
-  handleBinaryMessage(data) {
+  handleBinaryMessage(data, generation) {
+    if (generation !== void 0 && generation < this.currentGeneration) {
+      console.log(`\u{1F5D1}\uFE0F Discarding ghost audio (Gen ${generation} < ${this.currentGeneration})`);
+      return;
+    }
     if (this.audioManager) {
       this.audioManager.playAudio(data);
     }
@@ -647,7 +657,7 @@ var VoiceAgentClient = class {
         case "audio":
           if (msg.data) {
             const buffer = base64ToUint8Array(msg.data);
-            this.handleBinaryMessage(buffer);
+            this.handleBinaryMessage(buffer, msg.generation);
           }
           break;
         case "transcript":
@@ -666,6 +676,14 @@ var VoiceAgentClient = class {
           }
           break;
         case "status":
+          if (msg.data === "thinking") {
+            const newGen = msg.generation || 0;
+            if (newGen > this.currentGeneration) {
+              console.log(`\u{1F9E0} New thought (Gen ${newGen}) - Clearing audio queue`);
+              this.currentGeneration = newGen;
+              if (this.audioManager) this.audioManager.stopPlayback();
+            }
+          }
           if (msg.data === "interrupted" && this.audioManager) {
             this.audioManager.stopPlayback();
           }
@@ -687,6 +705,9 @@ var VoiceAgentClient = class {
           if (this.onError) this.onError(msg.data);
           console.error(`\u274C Server error: ${msg.data}`);
           break;
+        case "tool_call":
+          console.log(`\u{1F6E0}\uFE0F Tool Call: ${msg.name}(${msg.arguments})`);
+          break;
       }
     } catch (e) {
     }

package/dist/index.mjs

@@ -462,6 +462,7 @@ var VoiceAgentClient = class {
   prompt;
   voice;
   language;
+  tools = [];
   // Callbacks
   onTranscription;
   onResponse;
@@ -475,6 +476,7 @@ var VoiceAgentClient = class {
   wantVisemes = false;
   audioManager = null;
   enableAudio = false;
+  currentGeneration = 0;
   // Connection resilience
   isUserDisconnect = false;
   reconnecting = false;
@@ -493,6 +495,7 @@ var VoiceAgentClient = class {
     this.onError = config.onError;
     this.wantVisemes = config.visemes || false;
     this.enableAudio = config.enableAudio ?? false;
+    this.tools = config.tools || [];
   }
   /**
    * Connect to the Lokutor Voice Agent server
@@ -573,7 +576,10 @@ var VoiceAgentClient = class {
     this.ws.send(JSON.stringify({ type: "voice", data: this.voice }));
     this.ws.send(JSON.stringify({ type: "language", data: this.language }));
     this.ws.send(JSON.stringify({ type: "visemes", data: this.wantVisemes }));
-    console.log(`\u2699\uFE0F Configured: voice=${this.voice}, language=${this.language}, visemes=${this.wantVisemes}`);
+    if (this.tools && this.tools.length > 0) {
+      this.ws.send(JSON.stringify({ type: "tools", data: this.tools }));
+    }
+    console.log(`\u2699\uFE0F Configured: voice=${this.voice}, language=${this.language}, visemes=${this.wantVisemes}, tools=${this.tools.length}`);
   }
   /**
    * Send raw PCM audio data to the server
@@ -587,7 +593,11 @@ var VoiceAgentClient = class {
   /**
    * Handle incoming binary data (audio response)
    */
-  handleBinaryMessage(data) {
+  handleBinaryMessage(data, generation) {
+    if (generation !== void 0 && generation < this.currentGeneration) {
+      console.log(`\u{1F5D1}\uFE0F Discarding ghost audio (Gen ${generation} < ${this.currentGeneration})`);
+      return;
+    }
     if (this.audioManager) {
       this.audioManager.playAudio(data);
     }
@@ -603,7 +613,7 @@ var VoiceAgentClient = class {
         case "audio":
           if (msg.data) {
             const buffer = base64ToUint8Array(msg.data);
-            this.handleBinaryMessage(buffer);
+            this.handleBinaryMessage(buffer, msg.generation);
           }
           break;
         case "transcript":
@@ -622,6 +632,14 @@ var VoiceAgentClient = class {
           }
           break;
         case "status":
+          if (msg.data === "thinking") {
+            const newGen = msg.generation || 0;
+            if (newGen > this.currentGeneration) {
+              console.log(`\u{1F9E0} New thought (Gen ${newGen}) - Clearing audio queue`);
+              this.currentGeneration = newGen;
+              if (this.audioManager) this.audioManager.stopPlayback();
+            }
+          }
           if (msg.data === "interrupted" && this.audioManager) {
             this.audioManager.stopPlayback();
           }
@@ -643,6 +661,9 @@ var VoiceAgentClient = class {
           if (this.onError) this.onError(msg.data);
           console.error(`\u274C Server error: ${msg.data}`);
           break;
+        case "tool_call":
+          console.log(`\u{1F6E0}\uFE0F Tool Call: ${msg.name}(${msg.arguments})`);
+          break;
       }
     } catch (e) {
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lokutor/sdk",
-  "version": "1.1.11",
+  "version": "1.1.12",
   "description": "JavaScript/TypeScript SDK for Lokutor Real-time Voice AI",
   "main": "./dist/index.js",
   "module": "./dist/index.mjs",