Merge pull request #29 from AssemblyAI/5FC74E4F61B92B08B5515C2BF0B00557

Release 4.1.0

Author: Swimburger

CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## [4.1.0] - 2023-12-22
+
+### Added
+
+- Add `"anthropic/claude-2-1"` to `LemurModel` type
+- Add `encoding` option to the real-time service and factory. `encoding` can be `"pcm_s16le"` or `"pcm_mulaw"`.
+- `"pcm_mulaw"` is a newly supported audio encoding for the real-time service.
+
+### Changed
+
+- Allow any string into `final_model` for LeMUR requests
+
 ## [4.0.1] - 2023-12-14
 
 ### Added

package.json

@@ -1,6 +1,6 @@
 {
   "name": "assemblyai",
-  "version": "4.0.1",
+  "version": "4.1.0",
   "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"

scripts/generate-types.ts

@@ -17,6 +17,11 @@ async function generateTypes(apiSpecPath: string, outputPath: string) {
         return schemaObject.nullable ? "Date | null" : "Date";
       }
     },
+    postTransform(type) {
+      if (type === `components["schemas"]["LemurModel"] | string`) {
+        return `LiteralUnion<components["schemas"]["LemurModel"], string>`;
+      }
+    },
   });
   const schemasPosition = output.indexOf("schemas: {") + 10;
   output = output
@@ -46,6 +51,8 @@ async function generateTypes(apiSpecPath: string, outputPath: string) {
 /* tslint:disable */
 /* eslint-disable */
 
+import { LiteralUnion } from "./helpers";
+
 /** OneOf type helpers */
 type Without<T, U> = { [P in Exclude<keyof T, keyof U>]?: never };
 type XOR<T, U> = (T | U) extends object ? (Without<T, U> & U) | (Without<U, T> & T) : T | U;

scripts/kitchensink.ts

@@ -12,21 +12,20 @@ import {
 } from "../src";
 
 const client = new AssemblyAI({
-  apiKey: process.env.ASSEMBLYAI_API_KEY || "",
+  apiKey: process.env.ASSEMBLYAI_API_KEY!,
 });
 
 (async function transcribeUsingRealtime() {
   const useToken = false;
-  let token: undefined | string = undefined;
-  if (useToken) {
-    token = await client.realtime.createTemporaryToken({
-      expires_in: 480,
-    });
-  }
   const serviceParams: CreateRealtimeServiceParams = {
     sampleRate: 16_000,
     wordBoost: ["gore", "climate"],
-    token: token,
+    token: useToken
+      ? await client.realtime.createTemporaryToken({
+          expires_in: 480,
+        })
+      : undefined,
+    encoding: "pcm_s16le",
   };
   const rt = client.realtime.createService(serviceParams);
 
@@ -51,7 +50,7 @@ const client = new AssemblyAI({
     await rt.connect();
 
     const chunkSize = 8 * 1024;
-    const audio = createReadStream("./tests/static/gore-short.wav", {
+    const audio = createReadStream("./tests/static/gore.wav", {
       highWaterMark: chunkSize,
     });
     for await (const chunk of audio) {

src/services/realtime/factory.ts

@@ -16,12 +16,12 @@ export class RealtimeServiceFactory extends BaseService {
   }
 
   createService(params?: CreateRealtimeServiceParams): RealtimeService {
-    if (!params) params = { apiKey: this.rtFactoryParams.apiKey };
-    else if (!("token" in params) && !params.apiKey) {
-      params.apiKey = this.rtFactoryParams.apiKey;
+    const serviceParams = { ...params } as Record<string, unknown>;
+    if (!serviceParams.token && !serviceParams.apiKey) {
+      serviceParams.apiKey = this.rtFactoryParams.apiKey;
     }
 
-    return new RealtimeService(params as RealtimeServiceParams);
+    return new RealtimeService(serviceParams as RealtimeServiceParams);
   }
 
   async createTemporaryToken(params: RealtimeTokenParams) {

src/services/realtime/service.ts

@@ -10,6 +10,7 @@ import {
   PartialTranscript,
   FinalTranscript,
   SessionBeginsEventData,
+  AudioEncoding,
 } from "../..";
 import {
   RealtimeError,
@@ -23,6 +24,7 @@ export class RealtimeService {
   private realtimeUrl: string;
   private sampleRate: number;
   private wordBoost?: string[];
+  private encoding?: AudioEncoding;
   private apiKey?: string;
   private token?: string;
   private socket?: WebSocket;
@@ -33,10 +35,11 @@ export class RealtimeService {
     this.realtimeUrl = params.realtimeUrl ?? defaultRealtimeUrl;
     this.sampleRate = params.sampleRate ?? 16_000;
     this.wordBoost = params.wordBoost;
-    if ("apiKey" in params && params.apiKey) this.apiKey = params.apiKey;
+    this.encoding = params.encoding;
     if ("token" in params && params.token) this.token = params.token;
+    if ("apiKey" in params && params.apiKey) this.apiKey = params.apiKey;
 
-    if (!(this.apiKey || this.token)) {
+    if (!(this.token || this.apiKey)) {
       throw new Error("API key or temporary token is required.");
     }
   }
@@ -56,6 +59,9 @@ export class RealtimeService {
     if (this.wordBoost && this.wordBoost.length > 0) {
       searchParams.set("word_boost", JSON.stringify(this.wordBoost));
     }
+    if (this.encoding) {
+      searchParams.set("encoding", this.encoding);
+    }
     url.search = searchParams.toString();
 
     return url;

src/types/asyncapi.generated.ts

@@ -2,6 +2,8 @@
 /* tslint:disable */
 /* eslint-disable */
 
+import { LiteralUnion } from "./helpers";
+
 /** OneOf type helpers */
 type Without<T, U> = { [P in Exclude<keyof T, keyof U>]?: never };
 type XOR<T, U> = T | U extends object
@@ -18,6 +20,13 @@ export type AudioData = {
   audio_data: string;
 };
 
+/**
+ * @description The encoding of the audio data
+ * @default pcm_s16le
+ * @enum {string}
+ */
+export type AudioEncoding = "pcm_s16le" | "pcm_mulaw";
+
 export type FinalTranscript = RealtimeBaseTranscript & {
   /**
    * @description Describes the type of message

src/types/helpers/index.ts

@@ -0,0 +1,4 @@
+// source: https://github.com/sindresorhus/type-fest/blob/main/source/literal-union.d.ts
+export type LiteralUnion<LiteralType, BaseType> =
+  | LiteralType
+  | (BaseType & Record<never, never>);

src/types/openapi.generated.ts

@@ -2,6 +2,8 @@
 /* tslint:disable */
 /* eslint-disable */
 
+import { LiteralUnion } from "./helpers";
+
 /** OneOf type helpers */
 type Without<T, U> = { [P in Exclude<keyof T, keyof U>]?: never };
 type XOR<T, U> = T | U extends object
@@ -512,7 +514,7 @@ export type LemurBaseParams = {
       }
     ]
   >;
-  final_model?: LemurModel;
+  final_model?: LiteralUnion<LemurModel, string>;
   /**
    * @description Custom formatted transcript data. Maximum size is the context limit of the selected model, which defaults to 100000.
    * Use either transcript_ids or input_text as input into LeMUR.
@@ -552,7 +554,11 @@ export type LemurBaseResponse = {
  *
  * @enum {string}
  */
-export type LemurModel = "default" | "basic" | "assemblyai/mistral-7b";
+export type LemurModel =
+  | "default"
+  | "basic"
+  | "assemblyai/mistral-7b"
+  | "anthropic/claude-2-1";
 
 /**
  * @example {

src/types/realtime/index.ts

@@ -1,4 +1,5 @@
 import {
+  AudioEncoding,
   FinalTranscript,
   PartialTranscript,
   RealtimeTranscript,
@@ -9,6 +10,7 @@ type CreateRealtimeServiceParams = {
   realtimeUrl?: string;
   sampleRate?: number;
   wordBoost?: string[];
+  encoding?: AudioEncoding;
 } & (
   | {
       apiKey?: string;
@@ -22,6 +24,7 @@ type RealtimeServiceParams = {
   realtimeUrl?: string;
   sampleRate?: number;
   wordBoost?: string[];
+  encoding?: AudioEncoding;
 } & (
   | {
       apiKey: string;