@drawdream/livespeech 0.1.14 → 0.1.16

package/README.md

@@ -154,6 +154,22 @@ client.audioStart();
 
 > **Note:** All other SDK methods and events work identically in both modes. The only code change is adding `pipelineMode: 'composed'` to your session config.
 
+### Event Correlation (`turnId`)
+
+In Composed mode, all events include a `turnId` field (monotonic counter starting from 0). Events sharing the same `turnId` belong to the same speech turn — use this to match `userTranscript`, `response`, `audio`, and `turnComplete` events together. In Live mode, `turnId` is not present.
+
+```typescript
+client.on('userTranscript', (e) => {
+    console.log(`Turn ${e.turnId}: User said '${e.text}'`);
+});
+client.on('response', (e) => {
+    if (e.isFinal) console.log(`Turn ${e.turnId}: AI responded '${e.text}'`);
+});
+client.on('turnComplete', (e) => {
+    console.log(`Turn ${e.turnId} complete`);
+});
+```
+
 ---
 
 # Advanced API
@@ -304,6 +320,7 @@ client.audioStart();  // AI speaks immediately
 |--------|---------|-------------|
 | `prePrompt` | - | System prompt |
 | `language` | `'en-US'` | Language code |
+| `outputLanguage` | - | TTS voice language override (composed mode only) |
 | `pipelineMode` | `'live'` | `'live'` (~300ms) or `'composed'` (~1-2s) |
 | `aiSpeaksFirst` | `false` | AI initiates (live mode only) |
 | `allowHarmCategory` | `false` | Disable safety filters |

package/dist/index.d.mts

@@ -222,6 +222,10 @@ interface SessionConfig {
      * Session duration configuration (enables duration limits when set)
      */
     sessionDuration?: SessionDurationConfig;
+    /**
+     * Enable raw audio archival for QA (default: false)
+     */
+    archiveAudio?: boolean;
 }
 /**
  * Internal resolved configuration with defaults applied
@@ -308,6 +312,8 @@ interface ReadyEvent {
 interface UserTranscriptEvent {
     type: 'userTranscript';
     text: string;
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -317,6 +323,8 @@ interface ResponseEvent {
     type: 'response';
     text: string;
     isFinal: boolean;
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -327,6 +335,8 @@ interface AudioEvent {
     data: Uint8Array;
     format: string;
     sampleRate: number;
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -359,6 +369,8 @@ interface ReconnectingEvent {
  */
 interface TurnCompleteEvent {
     type: 'turnComplete';
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -473,6 +485,7 @@ interface StartSessionMessage extends BaseClientMessage {
     sessionMaxDurationSeconds?: number;
     enableSessionWarning?: boolean;
     enableSessionGoodbye?: boolean;
+    archiveAudio?: boolean;
 }
 /**
  * End session message
@@ -596,6 +609,7 @@ interface ServerSessionGoodbyeMessage extends BaseServerMessage {
 interface ServerUserTranscriptMessage extends BaseServerMessage {
     type: 'userTranscript';
     text: string;
+    turnId?: number;
 }
 /**
  * Response message from server
@@ -604,6 +618,7 @@ interface ServerResponseMessage extends BaseServerMessage {
     type: 'response';
     text: string;
     isFinal: boolean;
+    turnId?: number;
 }
 /**
  * Audio message from server
@@ -613,6 +628,7 @@ interface ServerAudioMessage extends BaseServerMessage {
     data: string;
     format: string;
     sampleRate: number;
+    turnId?: number;
 }
 /**
  * Error message from server
@@ -634,6 +650,7 @@ interface ServerPongMessage extends BaseServerMessage {
  */
 interface ServerTurnCompleteMessage extends BaseServerMessage {
     type: 'turnComplete';
+    turnId?: number;
 }
 /**
  * Ready message from server

package/dist/index.d.ts

@@ -222,6 +222,10 @@ interface SessionConfig {
      * Session duration configuration (enables duration limits when set)
      */
     sessionDuration?: SessionDurationConfig;
+    /**
+     * Enable raw audio archival for QA (default: false)
+     */
+    archiveAudio?: boolean;
 }
 /**
  * Internal resolved configuration with defaults applied
@@ -308,6 +312,8 @@ interface ReadyEvent {
 interface UserTranscriptEvent {
     type: 'userTranscript';
     text: string;
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -317,6 +323,8 @@ interface ResponseEvent {
     type: 'response';
     text: string;
     isFinal: boolean;
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -327,6 +335,8 @@ interface AudioEvent {
     data: Uint8Array;
     format: string;
     sampleRate: number;
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -359,6 +369,8 @@ interface ReconnectingEvent {
  */
 interface TurnCompleteEvent {
     type: 'turnComplete';
+    /** Monotonic turn counter for event correlation. Only present in Composed mode; `undefined` in Live mode. */
+    turnId?: number;
     timestamp: string;
 }
 /**
@@ -473,6 +485,7 @@ interface StartSessionMessage extends BaseClientMessage {
     sessionMaxDurationSeconds?: number;
     enableSessionWarning?: boolean;
     enableSessionGoodbye?: boolean;
+    archiveAudio?: boolean;
 }
 /**
  * End session message
@@ -596,6 +609,7 @@ interface ServerSessionGoodbyeMessage extends BaseServerMessage {
 interface ServerUserTranscriptMessage extends BaseServerMessage {
     type: 'userTranscript';
     text: string;
+    turnId?: number;
 }
 /**
  * Response message from server
@@ -604,6 +618,7 @@ interface ServerResponseMessage extends BaseServerMessage {
     type: 'response';
     text: string;
     isFinal: boolean;
+    turnId?: number;
 }
 /**
  * Audio message from server
@@ -613,6 +628,7 @@ interface ServerAudioMessage extends BaseServerMessage {
     data: string;
     format: string;
     sampleRate: number;
+    turnId?: number;
 }
 /**
  * Error message from server
@@ -634,6 +650,7 @@ interface ServerPongMessage extends BaseServerMessage {
  */
 interface ServerTurnCompleteMessage extends BaseServerMessage {
     type: 'turnComplete';
+    turnId?: number;
 }
 /**
  * Ready message from server

package/dist/index.js

@@ -738,6 +738,7 @@ var LiveSpeechClient = class {
         pipelineMode: config?.pipelineMode ?? "live",
         ...config?.aiSpeaksFirst && { aiSpeaksFirst: config.aiSpeaksFirst },
         allowHarmCategory: config?.allowHarmCategory ?? false,
+        archiveAudio: config?.archiveAudio ?? false,
         ...config?.tools && config.tools.length > 0 && { tools: config.tools },
         ...hasSessionDuration && {
           sessionMaxDurationSeconds: sessionDuration.maxSeconds,
@@ -1025,6 +1026,8 @@ var LiveSpeechClient = class {
     }
   }
   handleConnected(connectionId) {
+    this.sessionId = null;
+    this.isStreaming = false;
     const event = {
       type: "connected",
       connectionId,
@@ -1114,6 +1117,7 @@ var LiveSpeechClient = class {
           type: "response",
           text: message.text,
           isFinal: message.isFinal,
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("response", responseEvent);
@@ -1127,6 +1131,7 @@ var LiveSpeechClient = class {
           data: audioData,
           format: message.format,
           sampleRate: message.sampleRate,
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("audio", audioEvent);
@@ -1137,6 +1142,7 @@ var LiveSpeechClient = class {
         const userTranscriptEvent = {
           type: "userTranscript",
           text: message.text,
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("userTranscript", userTranscriptEvent);
@@ -1146,6 +1152,7 @@ var LiveSpeechClient = class {
       case "turnComplete": {
         const turnCompleteEvent = {
           type: "turnComplete",
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("turnComplete", turnCompleteEvent);

package/dist/index.mjs

@@ -699,6 +699,7 @@ var LiveSpeechClient = class {
         pipelineMode: config?.pipelineMode ?? "live",
         ...config?.aiSpeaksFirst && { aiSpeaksFirst: config.aiSpeaksFirst },
         allowHarmCategory: config?.allowHarmCategory ?? false,
+        archiveAudio: config?.archiveAudio ?? false,
         ...config?.tools && config.tools.length > 0 && { tools: config.tools },
         ...hasSessionDuration && {
           sessionMaxDurationSeconds: sessionDuration.maxSeconds,
@@ -986,6 +987,8 @@ var LiveSpeechClient = class {
     }
   }
   handleConnected(connectionId) {
+    this.sessionId = null;
+    this.isStreaming = false;
     const event = {
       type: "connected",
       connectionId,
@@ -1075,6 +1078,7 @@ var LiveSpeechClient = class {
           type: "response",
           text: message.text,
           isFinal: message.isFinal,
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("response", responseEvent);
@@ -1088,6 +1092,7 @@ var LiveSpeechClient = class {
           data: audioData,
           format: message.format,
           sampleRate: message.sampleRate,
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("audio", audioEvent);
@@ -1098,6 +1103,7 @@ var LiveSpeechClient = class {
         const userTranscriptEvent = {
           type: "userTranscript",
           text: message.text,
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("userTranscript", userTranscriptEvent);
@@ -1107,6 +1113,7 @@ var LiveSpeechClient = class {
       case "turnComplete": {
         const turnCompleteEvent = {
           type: "turnComplete",
+          ...message.turnId != null && { turnId: message.turnId },
           timestamp: message.timestamp
         };
         this.emit("turnComplete", turnCompleteEvent);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@drawdream/livespeech",
-  "version": "0.1.14",
+  "version": "0.1.16",
   "description": "Real-time speech-to-speech AI conversation SDK",
   "main": "dist/index.js",
   "module": "dist/index.mjs",