assemblyai 4.34.0 → 4.34.4

package/src/services/streaming/service.ts

@@ -15,6 +15,7 @@ import {
   StreamingEventMessage,
   TurnEvent,
   LLMGatewayResponseEvent,
+  SpeakerRevisionEvent,
   StreamingUpdateConfiguration,
   StreamingForceEndpoint,
   WarningEvent,
@@ -262,6 +263,10 @@ export class StreamingTranscriber {
       searchParams.set("prompt", this.params.prompt);
     }
 
+    if (this.params.agentContext) {
+      searchParams.set("agent_context", this.params.agentContext);
+    }
+
     if (this.params.filterProfanity) {
       searchParams.set(
         "filter_profanity",
@@ -274,7 +279,9 @@ export class StreamingTranscriber {
         "[Deprecation Warning] The speech model `u3-pro` is deprecated and will be removed in a future release. Please use `u3-rt-pro` instead.",
       );
     }
-    searchParams.set("speech_model", this.params.speechModel.toString());
+    if (this.params.speechModel !== undefined) {
+      searchParams.set("speech_model", this.params.speechModel.toString());
+    }
 
     if (this.params.languageDetection !== undefined) {
       searchParams.set(
@@ -389,6 +396,10 @@ export class StreamingTranscriber {
       searchParams.set("redact_pii_sub", this.params.redactPiiSub);
     }
 
+    if (this.params.mode !== undefined) {
+      searchParams.set("mode", this.params.mode);
+    }
+
     if (this.params.llmGateway !== undefined) {
       searchParams.set("llm_gateway", JSON.stringify(this.params.llmGateway));
     }
@@ -404,6 +415,10 @@ export class StreamingTranscriber {
     event: "llmGatewayResponse",
     listener: (event: LLMGatewayResponseEvent) => void,
   ): void;
+  on(
+    event: "speakerRevision",
+    listener: (event: SpeakerRevisionEvent) => void,
+  ): void;
   on(event: "warning", listener: (event: WarningEvent) => void): void;
   on(event: "vad", listener: (event: VadFrame) => void): void;
   on(event: "error", listener: (error: Error) => void): void;
@@ -508,6 +523,10 @@ Learn more at https://github.com/AssemblyAI/assemblyai-node-sdk/blob/main/docs/c
             this.listeners.llmGatewayResponse?.(message);
             break;
           }
+          case "SpeakerRevision": {
+            this.listeners.speakerRevision?.(message);
+            break;
+          }
           case "Warning": {
             const warning = message as WarningEvent;
             console.warn(

package/src/types/streaming/index.ts

@@ -91,7 +91,8 @@ export type StreamingTranscriberParams = {
   keyterms?: string[];
   keytermsPrompt?: string[];
   prompt?: string;
-  speechModel: StreamingSpeechModel;
+  agentContext?: string;
+  speechModel?: StreamingSpeechModel;
   languageDetection?: boolean;
   domain?: StreamingDomain;
   inactivityTimeout?: number;
@@ -107,6 +108,7 @@ export type StreamingTranscriberParams = {
   redactPii?: boolean;
   redactPiiPolicies?: StreamingPiiPolicy[];
   redactPiiSub?: StreamingPiiSubstitution;
+  mode?: StreamingMode;
   llmGateway?: LLMGatewayConfig;
   webhookUrl?: string;
   webhookAuthHeaderName?: string;
@@ -146,6 +148,7 @@ export type StreamingEvents =
   | "turn"
   | "speechStarted"
   | "llmGatewayResponse"
+  | "speakerRevision"
   | "warning"
   | "vad"
   | "error";
@@ -156,6 +159,7 @@ export type StreamingListeners = {
   turn?: (event: TurnEvent) => void;
   speechStarted?: (event: SpeechStartedEvent) => void;
   llmGatewayResponse?: (event: LLMGatewayResponseEvent) => void;
+  speakerRevision?: (event: SpeakerRevisionEvent) => void;
   warning?: (event: WarningEvent) => void;
   vad?: (event: VadFrame) => void;
   error?: (error: Error) => void;
@@ -165,11 +169,14 @@ export type StreamingSpeechModel =
   | "universal-streaming-english"
   | "universal-streaming-multilingual"
   | "u3-rt-pro"
+  | "u3-rt-pro-beta-1"
   | "whisper-rt"
   | "u3-pro";
 
 export type StreamingDomain = "medical-v1";
 
+export type StreamingMode = "max_accuracy" | "min_latency" | "balanced";
+
 export type VoiceFocusModel = "near-field" | "far-field";
 
 export type StreamingPiiSubstitution = "hash" | "entity_name";
@@ -330,6 +337,7 @@ export type StreamingUpdateConfiguration = {
   format_turns?: boolean;
   keyterms_prompt?: string[];
   prompt?: string;
+  agent_context?: string;
   filter_profanity?: boolean;
   interruption_delay?: number;
   turn_left_pad_ms?: number;
@@ -358,12 +366,37 @@ export type LLMGatewayResponseEvent = {
   data: unknown;
 };
 
+/**
+ * A single earlier Turn whose speaker labels were revised by reclustering.
+ * Match by `turn_order` against the original Turn; replace its per-word
+ * `speaker` assignments (and the turn-level `speaker_label`) with these. Text
+ * and word timestamps are unchanged from the original Turn.
+ */
+export type SpeakerRevisionItem = {
+  turn_order: number;
+  speaker_label?: string;
+  words: StreamingWord[];
+};
+
+/**
+ * Server-side correction to previously-emitted Turns' speaker labels.
+ * Diarization-only (emitted only when `speakerLabels` is enabled). Sent once
+ * per offline-recluster resolve; `revisions` carries one entry per earlier
+ * Turn whose label actually changed (unchanged turns are omitted). Apply each
+ * entry by matching its `turn_order`.
+ */
+export type SpeakerRevisionEvent = {
+  type: "SpeakerRevision";
+  revisions: SpeakerRevisionItem[];
+};
+
 export type StreamingEventMessage =
   | BeginEvent
   | TurnEvent
   | SpeechStartedEvent
   | TerminationEvent
   | LLMGatewayResponseEvent
+  | SpeakerRevisionEvent
   | ErrorEvent
   | WarningEvent;