Merge pull request #116 from AssemblyAI/EA17A79D514289B819B9AE7C9AB1132F

Sync from internal repo (2025/11/05)

Author: he-james

package.json

@@ -1,6 +1,6 @@
 {
   "name": "assemblyai",
-  "version": "4.18.5",
+  "version": "4.19.0",
   "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

@@ -1656,6 +1656,20 @@ export type SpeakerOptions = {
   max_speakers_expected?: number | null;
 };
 
+/**
+ * Code switching language detection result
+ */
+export type CodeSwitchingLanguage = {
+  /**
+   * The language code detected
+   */
+  language_code: string;
+  /**
+   * The confidence score for this language detection, between 0.0 and 1.0
+   */
+  confidence: number;
+};
+
 /**
  * Options for controlling the behavior of Automatic Language Detection
  */
@@ -1668,6 +1682,14 @@ export type LanguageDetectionOptions = {
    * The language to fallback to in case the language detection does not predict any of the expected ones.
    */
   fallback_language?: string | null;
+  /**
+   * Should code switching be enabled for this transcription.
+   */
+  code_switching?: boolean | null;
+  /**
+   * The confidence threshold for the automatically detected code switching language.
+   */
+  code_switching_confidence_threshold?: number | null;
 };
 
 /**
@@ -2716,6 +2738,10 @@ export type Transcript = {
    * List of language codes detected in the audio file when language detection is enabled
    */
   language_codes: LiteralUnion<TranscriptLanguageCode, string>[] | null;
+  /**
+   * List of detected languages with confidence scores when code switching is enabled
+   */
+  code_switching_languages?: CodeSwitchingLanguage[] | null;
   /**
    * The confidence threshold for the automatically detected language.
    * An error will be returned if the language confidence is below this threshold.
@@ -3270,7 +3296,7 @@ export type TranscriptOptionalParams = {
   /**
    * Options for controlling the behavior of Automatic Language Detection
    */
-  language_detection_options?: LanguageDetectionOptions;
+  language_detection_options?: LanguageDetectionOptions | null;
   /**
    * Enable {@link https://www.assemblyai.com/docs/models/speech-recognition#multichannel-transcription | Multichannel } transcription, can be true or false.
    * @defaultValue false

tests/unit/language-detection-options.test.ts

@@ -0,0 +1,146 @@
+import fetchMock from "jest-fetch-mock";
+import { LanguageDetectionOptions } 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("language detection options", () => {
+  it("should create transcript with all language_detection_options", async () => {
+    const languageDetectionOptions: LanguageDetectionOptions = {
+      expected_languages: ["en", "es"],
+      fallback_language: "en",
+      code_switching: true,
+      code_switching_confidence_threshold: 0.8,
+    };
+
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      language_detection: true,
+      language_detection_options: languageDetectionOptions,
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+    expect(transcript.status).toBe("queued");
+
+    // Verify the request body included language_detection_options
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    expect(requestBody.language_detection).toBe(true);
+    expect(requestBody.language_detection_options).toEqual(
+      languageDetectionOptions,
+    );
+  });
+
+  it("should create transcript with only code_switching enabled", async () => {
+    const languageDetectionOptions: LanguageDetectionOptions = {
+      code_switching: true,
+    };
+
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      language_detection: true,
+      language_detection_options: languageDetectionOptions,
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    expect(requestBody.language_detection_options.code_switching).toBe(true);
+    expect(
+      requestBody.language_detection_options
+        .code_switching_confidence_threshold,
+    ).toBeUndefined();
+  });
+
+  it("should create transcript with code_switching and confidence threshold", async () => {
+    const languageDetectionOptions: LanguageDetectionOptions = {
+      code_switching: true,
+      code_switching_confidence_threshold: 0.75,
+    };
+
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      language_detection: true,
+      language_detection_options: languageDetectionOptions,
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    expect(requestBody.language_detection_options.code_switching).toBe(true);
+    expect(
+      requestBody.language_detection_options
+        .code_switching_confidence_threshold,
+    ).toBe(0.75);
+  });
+
+  it("should create transcript with only confidence threshold", async () => {
+    const languageDetectionOptions: LanguageDetectionOptions = {
+      code_switching_confidence_threshold: 0.9,
+    };
+
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      language_detection: true,
+      language_detection_options: languageDetectionOptions,
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    expect(
+      requestBody.language_detection_options.code_switching,
+    ).toBeUndefined();
+    expect(
+      requestBody.language_detection_options
+        .code_switching_confidence_threshold,
+    ).toBe(0.9);
+  });
+
+  it("should handle null language_detection_options", async () => {
+    fetchMock.doMockOnceIf(
+      requestMatches({ url: "/v2/transcript", method: "POST" }),
+      JSON.stringify({ id: transcriptId, status: "queued" }),
+    );
+
+    const transcript = await assembly.transcripts.submit({
+      audio_url: remoteAudioURL,
+      language_detection: true,
+      language_detection_options: null,
+    });
+
+    expect(transcript.id).toBe(transcriptId);
+
+    const requestBody = JSON.parse(fetchMock.mock.calls[0][1]?.body as string);
+    expect(requestBody.language_detection_options).toBe(null);
+  });
+});