feat: add effort to SpeakerIdentificationRequest

The speaker identification API accepts an `effort` field ("low" | "medium",
default "low") to trade cost for quality on harder audio, but it was missing
from the SDK types, so setting it failed to type check.

Fixes #167

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: MLawlz

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## [4.36.6]
+
+- Add `effort` to `SpeakerIdentificationRequest` — `"low"` (default) or `"medium"`, matching the [speaker identification docs](https://www.assemblyai.com/docs/speech-understanding/speaker-identification#controlling-effort). The field was already accepted by the API but missing from the SDK types, so setting it failed to type check
+
 ## [4.36.4]
 
 - Add `aac` to the streaming `encoding` options — accepts an AAC stream in ADTS framing. Like `opus`/`ogg_opus`, AAC is self-describing, so `sampleRate` is optional for it (it remains required for PCM encodings and for dual-channel mode)

package.json

@@ -1,6 +1,6 @@
 {
   "name": "assemblyai",
-  "version": "4.36.5",
+  "version": "4.36.6",
   "description": "The AssemblyAI JavaScript SDK provides an easy-to-use interface for interacting with the AssemblyAI API, which supports async and real-time transcription, as well as the latest LeMUR models.",
   "engines": {
     "node": ">=18"

src/types/openapi.generated.ts

@@ -1508,6 +1508,11 @@ export type SeverityScoreSummary = {
  */
 export type SpeakerType = "role" | "name";
 
+/**
+ * Effort level for speaker identification
+ */
+export type SpeakerIdentificationEffort = "low" | "medium";
+
 /**
  * Speaker identification configuration for speech understanding
  */
@@ -1520,6 +1525,14 @@ export type SpeakerIdentificationRequest = {
    * Known speaker values (required when speaker_type is 'role')
    */
   known_values?: string[];
+  /**
+   * How much effort to spend identifying speakers. Use 'medium' for higher complexity
+   * audio, such as meetings with interruptions or transcripts where names aren't
+   * clearly stated, at a higher cost.
+   *
+   * @defaultValue "low"
+   */
+  effort?: SpeakerIdentificationEffort;
 };
 
 /**

tests/unit/speaker-identification.test.ts

@@ -0,0 +1,89 @@
+import fetchMock from "jest-fetch-mock";
+import {
+  SpeakerIdentificationEffort,
+  SpeakerIdentificationRequest,
+} from "../../src";
+import { createClient, requestMatches } from "./utils";
+
+fetchMock.enableMocks();
+
+const assembly = createClient();
+const transcriptId = "transcript_123";
+const remoteAudioURL = "https://assembly.ai/espn.m4a";
+
+beforeEach(() => {
+  jest.clearAllMocks();
+  fetchMock.resetMocks();
+  fetchMock.doMock();
+});
+
+describe("speaker identification", () => {
+  it("should create transcript with speaker_identification effort", async () => {
+    const speakerIdentification: SpeakerIdentificationRequest = {
+      speaker_type: "name",
+      effort: "medium",
+    };
+
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      speaker_labels: true,
+      speech_understanding: {
+        request: { speaker_identification: speakerIdentification },
+      },
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+    expect(transcript.status).toBe("queued");
+
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    expect(
+      requestBody.speech_understanding.request.speaker_identification,
+    ).toEqual(speakerIdentification);
+  });
+
+  it("should create transcript with speaker_identification without effort", async () => {
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      speaker_labels: true,
+      speech_understanding: {
+        request: {
+          speaker_identification: {
+            speaker_type: "role",
+            known_values: ["Agent", "Customer"],
+          },
+        },
+      },
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    const speakerIdentification =
+      requestBody.speech_understanding.request.speaker_identification;
+    expect(speakerIdentification.speaker_type).toBe("role");
+    expect(speakerIdentification.known_values).toEqual(["Agent", "Customer"]);
+    expect(speakerIdentification.effort).toBeUndefined();
+  });
+
+  it("should accept every documented effort value", () => {
+    const efforts: SpeakerIdentificationEffort[] = ["low", "medium"];
+
+    for (const effort of efforts) {
+      const speakerIdentification: SpeakerIdentificationRequest = {
+        speaker_type: "name",
+        effort,
+      };
+      expect(speakerIdentification.effort).toBe(effort);
+    }
+  });
+});