assemblyai 4.5.0 → 4.6.1

package/dist/streaming.mjs

@@ -84,7 +84,14 @@ class RealtimeError extends Error {
 const defaultRealtimeUrl = "wss://api.assemblyai.com/v2/realtime/ws";
 const forceEndOfUtteranceMessage = `{"force_end_utterance":true}`;
 const terminateSessionMessage = `{"terminate_session":true}`;
+/**
+ * RealtimeTranscriber connects to the Streaming Speech-to-Text API and lets you transcribe audio in real-time.
+ */
 class RealtimeTranscriber {
+    /**
+     * Create a new RealtimeTranscriber.
+     * @param params - Parameters to configure the RealtimeTranscriber
+     */
     constructor(params) {
         var _a, _b;
         this.listeners = {};
@@ -125,10 +132,19 @@ class RealtimeTranscriber {
         url.search = searchParams.toString();
         return url;
     }
+    /**
+     * Add a listener for an event.
+     * @param event - The event to listen for.
+     * @param listener - The function to call when the event is emitted.
+     */
     // eslint-disable-next-line @typescript-eslint/no-explicit-any
     on(event, listener) {
         this.listeners[event] = listener;
     }
+    /**
+     * Connect to the server and begin a new session.
+     * @returns A promise that resolves when the connection is established and the session begins.
+     */
     connect() {
         return new Promise((resolve) => {
             if (this.socket) {
@@ -210,9 +226,17 @@ class RealtimeTranscriber {
             };
         });
     }
+    /**
+     * Send audio data to the server.
+     * @param audio - The audio data to send to the server.
+     */
     sendAudio(audio) {
         this.send(audio);
     }
+    /**
+     * Create a writable stream that can be used to send audio data to the server.
+     * @returns A writable stream that can be used to send audio data to the server.
+     */
     stream() {
         return new WritableStream({
             write: (chunk) => {
@@ -240,6 +264,11 @@ class RealtimeTranscriber {
         }
         this.socket.send(data);
     }
+    /**
+     * Close the connection to the server.
+     * @param waitForSessionTermination - If true, the method will wait for the session to be terminated before closing the connection.
+     * While waiting for the session to be terminated, you will receive the final transcript and session information.
+     */
     close() {
         return __awaiter(this, arguments, void 0, function* (waitForSessionTermination = true) {
             var _a;

package/dist/types/openapi.generated.d.ts

@@ -529,12 +529,7 @@ export type LemurActionItemsParams = LemurBaseParams & {
  * }
  * ```
  */
-export type LemurActionItemsResponse = LemurBaseResponse & {
-    /**
-     * The response generated by LeMUR
-     */
-    response: string;
-};
+export type LemurActionItemsResponse = LemurStringResponse;
 /**
  * @example
  * ```js
@@ -562,7 +557,6 @@ export type LemurBaseParams = {
     ]>;
     /**
      * The model that is used for the final prompt after compression is performed.
-     * Defaults to "default".
      *
      * @defaultValue "default
      */
@@ -614,7 +608,7 @@ export type LemurBaseResponse = {
  * The model that is used for the final prompt after compression is performed.
  *
  */
-export type LemurModel = "default" | "basic" | "assemblyai/mistral-7b" | "anthropic/claude-2-1";
+export type LemurModel = "anthropic/claude-3-5-sonnet" | "anthropic/claude-3-opus" | "anthropic/claude-3-haiku" | "anthropic/claude-3-sonnet" | "anthropic/claude-2-1" | "anthropic/claude-2" | "default" | "anthropic/claude-instant-1-2" | "basic" | "assemblyai/mistral-7b";
 /**
  * @example
  * ```js
@@ -732,7 +726,26 @@ export type LemurQuestionAnswerResponse = LemurBaseResponse & {
      */
     response: LemurQuestionAnswer[];
 };
-export type LemurResponse = LemurTaskResponse | LemurSummaryResponse | LemurQuestionAnswerResponse | LemurActionItemsResponse;
+export type LemurResponse = LemurStringResponse | LemurQuestionAnswerResponse;
+/**
+ * @example
+ * ```js
+ * {
+ *   "request_id": "5e1b27c2-691f-4414-8bc5-f14678442f9e",
+ *   "response": "Based on the transcript, the following locations were mentioned as being affected by wildfire smoke from Canada:\n\n- Maine\n- Maryland\n- Minnesota\n- Mid Atlantic region\n- Northeast region\n- New York City\n- Baltimore\n",
+ *   "usage": {
+ *     "input_tokens": 27,
+ *     "output_tokens": 3
+ *   }
+ * }
+ * ```
+ */
+export type LemurStringResponse = {
+    /**
+     * The response generated by LeMUR.
+     */
+    response: string;
+} & LemurBaseResponse;
 /**
  * @example
  * ```js
@@ -766,12 +779,7 @@ export type LemurSummaryParams = LemurBaseParams & {
  * }
  * ```
  */
-export type LemurSummaryResponse = LemurBaseResponse & {
-    /**
-     * The response generated by LeMUR
-     */
-    response: string;
-};
+export type LemurSummaryResponse = LemurStringResponse;
 /**
  * @example
  * ```js
@@ -806,12 +814,7 @@ export type LemurTaskParams = {
  * }
  * ```
  */
-export type LemurTaskResponse = {
-    /**
-     * The response generated by LeMUR.
-     */
-    response: string;
-} & LemurBaseResponse;
+export type LemurTaskResponse = LemurStringResponse;
 /**
  * The usage numbers for the LeMUR request
  */
@@ -1025,6 +1028,10 @@ export type RealtimeTemporaryTokenResponse = {
      */
     token: string;
 };
+/**
+ * The notification when the redacted audio is ready.
+ */
+export type RedactedAudioNotification = RedactedAudioResponse;
 /**
  * @example
  * ```js
@@ -1219,7 +1226,7 @@ export type SeverityScoreSummary = {
 /**
  * The speech model to use for the transcription.
  */
-export type SpeechModel = "best" | "nano" | "conformer-2";
+export type SpeechModel = "best" | "nano";
 /**
  * The replacement logic for detected PII, can be "entity_name" or "hash". See {@link https://www.assemblyai.com/docs/models/pii-redaction | PII redaction } for more details.
  */
@@ -2343,15 +2350,17 @@ export type Transcript = {
      */
     webhook_auth: boolean;
     /**
-     * The header name which should be sent back with webhook calls
+     * The header name to be sent with the transcript completed or failed webhook requests
      */
     webhook_auth_header_name?: string | null;
     /**
-     * The status code we received from your server when delivering your webhook, if a webhook URL was provided
+     * The status code we received from your server when delivering the transcript completed or failed webhook request, if a webhook URL was provided
      */
     webhook_status_code?: number | null;
     /**
-     * The URL to which we send webhooks upon transcription completion
+     * The URL to which we send webhook requests.
+     * We sends two different types of webhook requests.
+     * One request when a transcript is completed or failed, and one request when the redacted audio is ready if redact_pii_audio is enabled.
      */
     webhook_url?: string | null;
     /**
@@ -2660,17 +2669,17 @@ export type TranscriptOptionalParams = {
      */
     topics?: string[];
     /**
-     * The header name which should be sent back with webhook calls
+     * The header name to be sent with the transcript completed or failed webhook requests
      * @defaultValue null
      */
     webhook_auth_header_name?: string | null;
     /**
-     * Specify a header name and value to send back with a webhook call for added security
+     * The header value to send back with the transcript completed or failed webhook requests for added security
      * @defaultValue null
      */
     webhook_auth_header_value?: string | null;
     /**
-     * The URL to which AssemblyAI send webhooks upon transcription completion
+     * The URL to which we send webhook requests. We sends two different types of webhook requests. One request when a transcript is completed or failed, and one request when the redacted audio is ready if redact_pii_audio is enabled.
      */
     webhook_url?: string;
     /**
@@ -3013,6 +3022,10 @@ export type TranscriptUtterance = {
      */
     words: TranscriptWord[];
 };
+/**
+ * The notifications sent to the webhook URL.
+ */
+export type TranscriptWebhookNotification = TranscriptReadyNotification | RedactedAudioNotification;
 /**
  * @example
  * ```js

package/dist/workerd.mjs

@@ -15,7 +15,7 @@ if (typeof navigator !== "undefined" && navigator.userAgent) {
     defaultUserAgentString += navigator.userAgent;
 }
 const defaultUserAgent = {
-    sdk: { name: "JavaScript", version: "4.5.0" },
+    sdk: { name: "JavaScript", version: "4.6.1" },
 };
 if (typeof process !== "undefined") {
     if (process.versions.node && defaultUserAgentString.indexOf("Node") === -1) {
@@ -191,7 +191,14 @@ class RealtimeError extends Error {
 const defaultRealtimeUrl = "wss://api.assemblyai.com/v2/realtime/ws";
 const forceEndOfUtteranceMessage = `{"force_end_utterance":true}`;
 const terminateSessionMessage = `{"terminate_session":true}`;
+/**
+ * RealtimeTranscriber connects to the Streaming Speech-to-Text API and lets you transcribe audio in real-time.
+ */
 class RealtimeTranscriber {
+    /**
+     * Create a new RealtimeTranscriber.
+     * @param params - Parameters to configure the RealtimeTranscriber
+     */
     constructor(params) {
         this.listeners = {};
         this.realtimeUrl = params.realtimeUrl ?? defaultRealtimeUrl;
@@ -231,10 +238,19 @@ class RealtimeTranscriber {
         url.search = searchParams.toString();
         return url;
     }
+    /**
+     * Add a listener for an event.
+     * @param event - The event to listen for.
+     * @param listener - The function to call when the event is emitted.
+     */
     // eslint-disable-next-line @typescript-eslint/no-explicit-any
     on(event, listener) {
         this.listeners[event] = listener;
     }
+    /**
+     * Connect to the server and begin a new session.
+     * @returns A promise that resolves when the connection is established and the session begins.
+     */
     connect() {
         return new Promise((resolve) => {
             if (this.socket) {
@@ -313,9 +329,17 @@ class RealtimeTranscriber {
             };
         });
     }
+    /**
+     * Send audio data to the server.
+     * @param audio - The audio data to send to the server.
+     */
     sendAudio(audio) {
         this.send(audio);
     }
+    /**
+     * Create a writable stream that can be used to send audio data to the server.
+     * @returns A writable stream that can be used to send audio data to the server.
+     */
     stream() {
         return new WritableStream({
             write: (chunk) => {
@@ -343,6 +367,11 @@ class RealtimeTranscriber {
         }
         this.socket.send(data);
     }
+    /**
+     * Close the connection to the server.
+     * @param waitForSessionTermination - If true, the method will wait for the session to be terminated before closing the connection.
+     * While waiting for the session to be terminated, you will receive the final transcript and session information.
+     */
     async close(waitForSessionTermination = true) {
         if (this.socket) {
             if (this.socket.readyState === this.socket.OPEN) {
@@ -429,7 +458,6 @@ class TranscriptService extends BaseService {
      * @returns A promise that resolves to the transcript. The transcript status is "completed" or "error".
      */
     async transcribe(params, options) {
-        deprecateConformer2(params);
         const transcript = await this.submit(params);
         return await this.waitUntilReady(transcript.id, options);
     }
@@ -439,7 +467,6 @@ class TranscriptService extends BaseService {
      * @returns A promise that resolves to the queued transcript.
      */
     async submit(params) {
-        deprecateConformer2(params);
         let audioUrl;
         let transcriptParams = undefined;
         if ("audio" in params) {
@@ -483,7 +510,6 @@ class TranscriptService extends BaseService {
      * @deprecated Use `transcribe` instead to transcribe a audio file that includes polling, or `submit` to transcribe a audio file without polling.
      */
     async create(params, options) {
-        deprecateConformer2(params);
         const path = getPath(params.audio_url);
         if (path !== null) {
             const uploadUrl = await this.files.upload(path);
@@ -646,13 +672,6 @@ class TranscriptService extends BaseService {
         };
     }
 }
-function deprecateConformer2(params) {
-    if (!params)
-        return;
-    if (params.speech_model === "conformer-2") {
-        console.warn("The speech_model conformer-2 option is deprecated and will stop working in the near future. Use best or nano instead.");
-    }
-}
 
 const readFile = async function (
 // eslint-disable-next-line @typescript-eslint/no-unused-vars

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "assemblyai",
-  "version": "4.5.0",
+  "version": "4.6.1",
   "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"
@@ -115,39 +115,51 @@
     "docs"
   ],
   "devDependencies": {
-    "@babel/preset-env": "^7.24.5",
-    "@babel/preset-typescript": "^7.24.1",
+    "@babel/preset-env": "^7.24.7",
+    "@babel/preset-typescript": "^7.24.7",
     "@rollup/plugin-node-resolve": "^15.2.3",
-    "@rollup/plugin-replace": "^5.0.5",
+    "@rollup/plugin-replace": "^5.0.7",
     "@rollup/plugin-terser": "^0.4.4",
     "@rollup/plugin-typescript": "^11.1.6",
     "@types/jest": "^29.5.12",
-    "@types/node": "^18.19.32",
+    "@types/node": "^18.19.38",
     "@types/websocket": "^1.0.10",
     "@types/ws": "^8.5.10",
-    "@typescript-eslint/eslint-plugin": "^7.8.0",
+    "@typescript-eslint/eslint-plugin": "^7.13.1",
     "dotenv": "^16.4.5",
     "eslint": "^8.57.0",
-    "eslint-plugin-tsdoc": "^0.2.17",
+    "eslint-plugin-tsdoc": "^0.3.0",
     "jest": "^29.7.0",
     "jest-cli": "^29.7.0",
     "jest-fetch-mock": "^3.0.3",
     "jest-junit": "^16.0.0",
     "jest-websocket-mock": "^2.5.0",
-    "jsr": "^0.12.4",
     "mock-socket": "^9.3.1",
     "openapi-typescript": "^6.7.5",
-    "prettier": "^3.2.5",
-    "publint": "^0.2.7",
-    "rimraf": "^5.0.5",
-    "rollup": "^4.17.2",
-    "ts-jest": "^29.1.2",
-    "tslib": "^2.5.3",
+    "prettier": "^3.3.2",
+    "publint": "^0.2.8",
+    "rimraf": "^5.0.7",
+    "rollup": "^4.18.0",
+    "ts-jest": "^29.1.5",
+    "tslib": "^2.6.3",
     "typedoc": "^0.25.13",
     "typedoc-plugin-extras": "^3.0.0",
     "typescript": "^5.4.5"
   },
   "dependencies": {
-    "ws": "^8.17.0"
+    "ws": "^8.17.1"
+  },
+  "pnpm": {
+    "packageExtensions": {
+      "ws": {
+        "peerDependencies": {
+          "@types/ws": "^8.5.10"
+        }
+      }
+    },
+    "overrides": {
+      "undici@<5.28.4": ">=5.28.4",
+      "braces@<3.0.3": ">=3.0.3"
+    }
   }
 }

package/src/services/realtime/service.ts

@@ -45,6 +45,9 @@ type BufferLike =
   | { valueOf(): string }
   | { [Symbol.toPrimitive](hint: string): string };
 
+/**
+ * RealtimeTranscriber connects to the Streaming Speech-to-Text API and lets you transcribe audio in real-time.
+ */
 export class RealtimeTranscriber {
   private realtimeUrl: string;
   private sampleRate: number;
@@ -59,6 +62,10 @@ export class RealtimeTranscriber {
   private listeners: RealtimeListeners = {};
   private sessionTerminatedResolve?: () => void;
 
+  /**
+   * Create a new RealtimeTranscriber.
+   * @param params - Parameters to configure the RealtimeTranscriber
+   */
   constructor(params: RealtimeTranscriberParams) {
     this.realtimeUrl = params.realtimeUrl ?? defaultRealtimeUrl;
     this.sampleRate = params.sampleRate ?? 16_000;
@@ -106,30 +113,75 @@ export class RealtimeTranscriber {
     return url;
   }
 
+  /**
+   * Listen for the open event which is emitted when the connection is established and the session begins.
+   * @param event - The open event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(event: "open", listener: (event: SessionBeginsEventData) => void): void;
+  /**
+   * Listen for the transcript event which is emitted when a partian or final transcript is received.
+   * @param event - The transcript event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(
     event: "transcript",
     listener: (transcript: RealtimeTranscript) => void,
   ): void;
+  /**
+   * Listen for the partial transcript event which is emitted when a partial transcript is received.
+   * @param event - The partial transcript event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(
     event: "transcript.partial",
     listener: (transcript: PartialTranscript) => void,
   ): void;
+  /**
+   * Listen for the final transcript event which is emitted when a final transcript is received.
+   * @param event - The final transcript event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(
     event: "transcript.final",
     listener: (transcript: FinalTranscript) => void,
   ): void;
+  /**
+   * Listen for the session information event which is emitted when session information is received.
+   * The session information is sent right before the session is terminated.
+   * @param event - The session information event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(
     event: "session_information",
     listener: (info: SessionInformation) => void,
   ): void;
+  /**
+   * Listen for the error event which is emitted when an error occurs.
+   * @param event - The error event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(event: "error", listener: (error: Error) => void): void;
+  /**
+   * Listen for the close event which is emitted when the connection is closed.
+   * @param event - The close event.
+   * @param listener - The function to call when the event is emitted.
+   */
   on(event: "close", listener: (code: number, reason: string) => void): void;
+  /**
+   * Add a listener for an event.
+   * @param event - The event to listen for.
+   * @param listener - The function to call when the event is emitted.
+   */
   // eslint-disable-next-line @typescript-eslint/no-explicit-any
   on(event: RealtimeEvents, listener: (...args: any[]) => void) {
     this.listeners[event] = listener;
   }
 
+  /**
+   * Connect to the server and begin a new session.
+   * @returns A promise that resolves when the connection is established and the session begins.
+   */
   connect() {
     return new Promise<SessionBeginsEventData>((resolve) => {
       if (this.socket) {
@@ -216,10 +268,18 @@ export class RealtimeTranscriber {
     });
   }
 
+  /**
+   * Send audio data to the server.
+   * @param audio - The audio data to send to the server.
+   */
   sendAudio(audio: AudioData) {
     this.send(audio);
   }
 
+  /**
+   * Create a writable stream that can be used to send audio data to the server.
+   * @returns A writable stream that can be used to send audio data to the server.
+   */
   stream(): WritableStream<AudioData> {
     return new WritableStream<AudioData>({
       write: (chunk: AudioData) => {
@@ -251,6 +311,11 @@ export class RealtimeTranscriber {
     this.socket.send(data);
   }
 
+  /**
+   * Close the connection to the server.
+   * @param waitForSessionTermination - If true, the method will wait for the session to be terminated before closing the connection.
+   * While waiting for the session to be terminated, you will receive the final transcript and session information.
+   */
   async close(waitForSessionTermination = true) {
     if (this.socket) {
       if (this.socket.readyState === this.socket.OPEN) {

package/src/services/transcripts/index.ts

@@ -15,7 +15,6 @@ import {
   TranscribeParams,
   TranscribeOptions,
   SubmitParams,
-  SpeechModel,
   RedactedAudioFile,
 } from "../..";
 import { FileService } from "../files";
@@ -39,7 +38,6 @@ export class TranscriptService extends BaseService {
     params: TranscribeParams,
     options?: TranscribeOptions,
   ): Promise<Transcript> {
-    deprecateConformer2(params);
     const transcript = await this.submit(params);
     return await this.waitUntilReady(transcript.id, options);
   }
@@ -50,7 +48,6 @@ export class TranscriptService extends BaseService {
    * @returns A promise that resolves to the queued transcript.
    */
   async submit(params: SubmitParams): Promise<Transcript> {
-    deprecateConformer2(params);
     let audioUrl;
     let transcriptParams: TranscriptParams | undefined = undefined;
     if ("audio" in params) {
@@ -95,7 +92,6 @@ export class TranscriptService extends BaseService {
     params: TranscriptParams,
     options?: CreateTranscriptOptions,
   ): Promise<Transcript> {
-    deprecateConformer2(params);
     const path = getPath(params.audio_url);
     if (path !== null) {
       const uploadUrl = await this.files.upload(path);
@@ -287,12 +283,3 @@ export class TranscriptService extends BaseService {
     };
   }
 }
-
-function deprecateConformer2(params: { speech_model?: SpeechModel | null }) {
-  if (!params) return;
-  if (params.speech_model === "conformer-2") {
-    console.warn(
-      "The speech_model conformer-2 option is deprecated and will stop working in the near future. Use best or nano instead.",
-    );
-  }
-}