@cartesia/cartesia-js 0.0.4-alpha.0 → 1.0.0-alpha.2

package/dist/{audio → tts}/utils.cjs

@@ -27,14 +27,12 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
 ));
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 
-// src/audio/utils.ts
+// src/tts/utils.ts
 var utils_exports = {};
 __export(utils_exports, {
   base64ToArray: () => base64ToArray,
-  bufferToWav: () => bufferToWav,
   createMessageHandlerForContextId: () => createMessageHandlerForContextId,
   filterSentinel: () => filterSentinel,
-  getBufferDuration: () => getBufferDuration,
   getEmitteryCallbacks: () => getEmitteryCallbacks,
   getSentinel: () => getSentinel,
   isComplete: () => isComplete,
@@ -43,38 +41,39 @@ __export(utils_exports, {
 });
 module.exports = __toCommonJS(utils_exports);
 var import_base64_js = __toESM(require("base64-js"), 1);
-
-// src/lib/constants.ts
-var SAMPLE_RATE = 44100;
-
-// src/audio/utils.ts
-function getBufferDuration(b64) {
-  const floats = base64ToArray(b64);
-  return floats.length / SAMPLE_RATE;
-}
 function base64ToArray(b64) {
-  return filterSentinel(b64).reduce((acc, b) => {
-    const floats = new Float32Array(import_base64_js.default.toByteArray(b).buffer);
-    const newAcc = new Float32Array(acc.length + floats.length);
-    newAcc.set(acc, 0);
-    newAcc.set(floats, acc.length);
-    return newAcc;
-  }, new Float32Array(0));
+  const byteArrays = filterSentinel(b64).map((b) => import_base64_js.default.toByteArray(b));
+  const totalLength = byteArrays.reduce(
+    (acc, arr) => acc + arr.length / Float32Array.BYTES_PER_ELEMENT,
+    0
+  );
+  const result = new Float32Array(totalLength);
+  let offset = 0;
+  for (const arr of byteArrays) {
+    const floats = new Float32Array(arr.buffer);
+    result.set(floats, offset);
+    offset += floats.length;
+  }
+  return result;
 }
-function playAudioBuffer(b64, context, maybeStartAt = null, onEnded = null) {
-  const startAt = maybeStartAt != null ? maybeStartAt : context.currentTime;
-  const floats = base64ToArray(b64);
+function playAudioBuffer(floats, context, startAt, sampleRate) {
   const source = context.createBufferSource();
-  const buffer = context.createBuffer(1, floats.length, SAMPLE_RATE);
+  const buffer = context.createBuffer(1, floats.length, sampleRate);
   buffer.getChannelData(0).set(floats);
   source.buffer = buffer;
   source.connect(context.destination);
   source.start(startAt);
-  source.onended = onEnded;
-  return buffer.duration;
+  return new Promise((resolve) => {
+    source.onended = () => {
+      resolve();
+    };
+  });
 }
 function createMessageHandlerForContextId(contextId, handler) {
   return (event) => {
+    if (typeof event.data !== "string") {
+      return;
+    }
     const message = JSON.parse(event.data);
     if (message.context_id !== contextId) {
       return;
@@ -85,7 +84,7 @@ function createMessageHandlerForContextId(contextId, handler) {
     } else {
       chunk = message.data;
     }
-    handler({ chunk, message });
+    handler({ chunk, message: event.data });
   };
 }
 function getSentinel() {
@@ -110,45 +109,11 @@ function getEmitteryCallbacks(emitter) {
     events: emitter.events.bind(emitter)
   };
 }
-function bufferToWav(sampleRate, channelBuffers) {
-  const totalSamples = channelBuffers[0].length * channelBuffers.length;
-  const buffer = new ArrayBuffer(44 + totalSamples * 2);
-  const view = new DataView(buffer);
-  const writeString = (view2, offset2, string) => {
-    for (let i = 0; i < string.length; i++) {
-      view2.setUint8(offset2 + i, string.charCodeAt(i));
-    }
-  };
-  writeString(view, 0, "RIFF");
-  view.setUint32(4, 36 + totalSamples * 2, true);
-  writeString(view, 8, "WAVE");
-  writeString(view, 12, "fmt ");
-  view.setUint32(16, 16, true);
-  view.setUint16(20, 1, true);
-  view.setUint16(22, channelBuffers.length, true);
-  view.setUint32(24, sampleRate, true);
-  view.setUint32(28, sampleRate * 4, true);
-  view.setUint16(32, channelBuffers.length * 2, true);
-  view.setUint16(34, 16, true);
-  writeString(view, 36, "data");
-  view.setUint32(40, totalSamples * 2, true);
-  let offset = 44;
-  for (let i = 0; i < channelBuffers[0].length; i++) {
-    for (let channel = 0; channel < channelBuffers.length; channel++) {
-      const s = Math.max(-1, Math.min(1, channelBuffers[channel][i]));
-      view.setInt16(offset, s < 0 ? s * 32768 : s * 32767, true);
-      offset += 2;
-    }
-  }
-  return buffer;
-}
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   base64ToArray,
-  bufferToWav,
   createMessageHandlerForContextId,
   filterSentinel,
-  getBufferDuration,
   getEmitteryCallbacks,
   getSentinel,
   isComplete,

package/dist/tts/utils.d.cts

@@ -0,0 +1,67 @@
+import emittery__default from 'emittery';
+import { Chunk, Sentinel, EmitteryCallbacks } from '../types/index.cjs';
+
+/**
+ * Convert base64-encoded audio buffer(s) to a Float32Array.
+ *
+ * @param b64 The base64-encoded audio buffer, or an array of base64-encoded
+ * audio buffers.
+ * @returns The audio buffer(s) as a Float32Array.
+ */
+declare function base64ToArray(b64: Chunk[]): Float32Array;
+/**
+ * Schedule an audio buffer to play at a given time in the passed context.
+ *
+ * @param floats The audio buffer to play.
+ * @param context The audio context to play the buffer in.
+ * @param startAt The time to start playing the buffer at.
+ * @param sampleRate The sample rate of the audio.
+ * @returns A promise that resolves when the audio has finished playing.
+ */
+declare function playAudioBuffer(floats: Float32Array, context: AudioContext, startAt: number, sampleRate: number): Promise<void>;
+/**
+ * Unwraps a chunk of audio data from a message event and calls the
+ * handler with it if the context ID matches.
+ *
+ * @param contextId The context ID to listen for.
+ * @param handler The handler to call with the chunk of audio data.
+ * @returns A message event handler.
+ */
+declare function createMessageHandlerForContextId(contextId: string, handler: ({ chunk, message, }: {
+    chunk: Chunk;
+    message: string;
+}) => void): (event: MessageEvent) => void;
+/**
+ * Get a sentinel value that indicates the end of a stream.
+ * @returns A sentinel value to indicate the end of a stream.
+ */
+declare function getSentinel(): Sentinel;
+/**
+ * Check if a chunk is a sentinel value (i.e. null).
+ *
+ * @param chunk
+ * @returns Whether the chunk is a sentinel value.
+ */
+declare function isSentinel(x: unknown): x is Sentinel;
+/**
+ * Filter out null values from a collection.
+ *
+ * @param collection The collection to filter.
+ * @returns The collection with null values removed.
+ */
+declare function filterSentinel<T>(collection: T[]): Exclude<T, Sentinel>[];
+/**
+ * Check if an array of chunks is complete by testing if the last chunk is a sentinel
+ * value (i.e. null).
+ * @param chunk
+ * @returns Whether the array of chunks is complete.
+ */
+declare function isComplete(chunks: Chunk[]): boolean;
+/**
+ * Get user-facing emitter callbacks for an Emittery instance.
+ * @param emitter The Emittery instance to get callbacks for.
+ * @returns User-facing emitter callbacks.
+ */
+declare function getEmitteryCallbacks<T>(emitter: emittery__default<T>): EmitteryCallbacks<T>;
+
+export { base64ToArray, createMessageHandlerForContextId, filterSentinel, getEmitteryCallbacks, getSentinel, isComplete, isSentinel, playAudioBuffer };

package/dist/tts/utils.d.ts

@@ -0,0 +1,67 @@
+import emittery__default from 'emittery';
+import { Chunk, Sentinel, EmitteryCallbacks } from '../types/index.js';
+
+/**
+ * Convert base64-encoded audio buffer(s) to a Float32Array.
+ *
+ * @param b64 The base64-encoded audio buffer, or an array of base64-encoded
+ * audio buffers.
+ * @returns The audio buffer(s) as a Float32Array.
+ */
+declare function base64ToArray(b64: Chunk[]): Float32Array;
+/**
+ * Schedule an audio buffer to play at a given time in the passed context.
+ *
+ * @param floats The audio buffer to play.
+ * @param context The audio context to play the buffer in.
+ * @param startAt The time to start playing the buffer at.
+ * @param sampleRate The sample rate of the audio.
+ * @returns A promise that resolves when the audio has finished playing.
+ */
+declare function playAudioBuffer(floats: Float32Array, context: AudioContext, startAt: number, sampleRate: number): Promise<void>;
+/**
+ * Unwraps a chunk of audio data from a message event and calls the
+ * handler with it if the context ID matches.
+ *
+ * @param contextId The context ID to listen for.
+ * @param handler The handler to call with the chunk of audio data.
+ * @returns A message event handler.
+ */
+declare function createMessageHandlerForContextId(contextId: string, handler: ({ chunk, message, }: {
+    chunk: Chunk;
+    message: string;
+}) => void): (event: MessageEvent) => void;
+/**
+ * Get a sentinel value that indicates the end of a stream.
+ * @returns A sentinel value to indicate the end of a stream.
+ */
+declare function getSentinel(): Sentinel;
+/**
+ * Check if a chunk is a sentinel value (i.e. null).
+ *
+ * @param chunk
+ * @returns Whether the chunk is a sentinel value.
+ */
+declare function isSentinel(x: unknown): x is Sentinel;
+/**
+ * Filter out null values from a collection.
+ *
+ * @param collection The collection to filter.
+ * @returns The collection with null values removed.
+ */
+declare function filterSentinel<T>(collection: T[]): Exclude<T, Sentinel>[];
+/**
+ * Check if an array of chunks is complete by testing if the last chunk is a sentinel
+ * value (i.e. null).
+ * @param chunk
+ * @returns Whether the array of chunks is complete.
+ */
+declare function isComplete(chunks: Chunk[]): boolean;
+/**
+ * Get user-facing emitter callbacks for an Emittery instance.
+ * @param emitter The Emittery instance to get callbacks for.
+ * @returns User-facing emitter callbacks.
+ */
+declare function getEmitteryCallbacks<T>(emitter: emittery__default<T>): EmitteryCallbacks<T>;
+
+export { base64ToArray, createMessageHandlerForContextId, filterSentinel, getEmitteryCallbacks, getSentinel, isComplete, isSentinel, playAudioBuffer };

package/dist/{audio → tts}/utils.js

@@ -1,23 +1,18 @@
 import {
   base64ToArray,
-  bufferToWav,
   createMessageHandlerForContextId,
   filterSentinel,
-  getBufferDuration,
   getEmitteryCallbacks,
   getSentinel,
   isComplete,
   isSentinel,
   playAudioBuffer
-} from "../chunk-5TSWLYOW.js";
-import "../chunk-OVI3W3GG.js";
-import "../chunk-R4P7LWVZ.js";
+} from "../chunk-RO7TY474.js";
+import "../chunk-WIFMLPT5.js";
 export {
   base64ToArray,
-  bufferToWav,
   createMessageHandlerForContextId,
   filterSentinel,
-  getBufferDuration,
   getEmitteryCallbacks,
   getSentinel,
   isComplete,