@ssml-builder-js/azure-tts-client 2.12.0 → 2.13.0

package/src/types.ts

@@ -5,22 +5,32 @@ export interface TtsConfig {
   subscriptionKey: string;
   region: string;
   outputFormat?: string;
+  /** Original plain-text range represented by this synthesis request. */
+  sourceTextRange?: { start: number; end: number };
+  /** Reports completion of a chunk when using synthesizeSsmlChunks. */
+  onProgress?: (event: { currentChunk: number; totalChunks: number; percent: number }) => void;
 }
 
 export interface SsmlSynthesisBoundary {
   text: string;
   audioOffsetMs: number;
   durationMs: number;
+  textRange?: { start: number; end: number };
+  requestId?: string;
 }
 
 export interface SsmlSynthesisViseme {
   visemeId: number;
   audioOffsetMs: number;
+  textRange?: { start: number; end: number };
+  requestId?: string;
 }
 
 export interface SsmlSynthesisBookmark {
   name: string;
   audioOffsetMs: number;
+  textRange?: { start: number; end: number };
+  requestId?: string;
 }
 
 /** Audio and Azure Speech synchronization events emitted for one SSML request. */
@@ -34,6 +44,25 @@ export interface SsmlSynthesisResult {
   wordBoundaries?: SsmlSynthesisBoundary[];
   visemes?: SsmlSynthesisViseme[];
   bookmarks?: SsmlSynthesisBookmark[];
+  /** Request identifier returned by Azure Speech, when available. */
+  requestId?: string;
+  /** Original plain-text range represented by the result. */
+  textRange?: { start: number; end: number };
+}
+
+export interface SsmlSynthesisChunk {
+  ssml: string;
+  originalTextRange?: { start: number; end: number };
+}
+
+export interface SynthesizeChunksOptions {
+  onProgress?: (event: SynthesisProgressEvent) => void;
+}
+
+export interface SynthesisProgressEvent {
+  currentChunk: number;
+  totalChunks: number;
+  percent: number;
 }
 
 export interface AzureTtsLogger {
@@ -51,4 +80,5 @@ export interface AzureTtsClientOptions {
   endpoint?: string;
   outputFormat?: string;
   logger?: AzureTtsLogger;
+  onProgress?: (event: SynthesisProgressEvent) => void;
 }

package/test/v213-pipeline.test.ts

@@ -0,0 +1,68 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { mergeSynthesisResults, synthesizeSsmlSafe } from "../src/index.ts";
+
+const audio = (values: number[]): ArrayBuffer => Uint8Array.from(values).buffer;
+
+test("mergeSynthesisResults concatenates audio and offsets synchronization events", () => {
+  const result = mergeSynthesisResults([
+    {
+      audioData: audio([1, 2]),
+      durationMs: 100,
+      boundaries: [{ text: "one", audioOffsetMs: 20, durationMs: 30, textRange: { start: 0, end: 3 }, requestId: "a" }],
+      visemes: [{ visemeId: 1, audioOffsetMs: 40 }],
+      bookmarks: [{ name: "first", audioOffsetMs: 50 }],
+    },
+    {
+      audioData: audio([3, 4, 5]),
+      durationMs: 250,
+      boundaries: [{ text: "two", audioOffsetMs: 10, durationMs: 20, textRange: { start: 3, end: 6 }, requestId: "b" }],
+      visemes: [{ visemeId: 2, audioOffsetMs: 15 }],
+      bookmarks: [{ name: "second", audioOffsetMs: 25 }],
+    },
+  ]);
+
+  assert.deepEqual([...new Uint8Array(result.audioData)], [1, 2, 3, 4, 5]);
+  assert.equal(result.durationMs, 350);
+  assert.deepEqual(
+    result.boundaries?.map(({ audioOffsetMs }) => audioOffsetMs),
+    [20, 110],
+  );
+  assert.deepEqual(
+    result.visemes?.map(({ audioOffsetMs }) => audioOffsetMs),
+    [40, 115],
+  );
+  assert.deepEqual(
+    result.bookmarks?.map(({ audioOffsetMs }) => audioOffsetMs),
+    [50, 125],
+  );
+  assert.deepEqual(result.boundaries?.[1]?.textRange, { start: 3, end: 6 });
+});
+
+test("synthesizeSsmlSafe blocks invalid SSML without calling the client", async () => {
+  let calls = 0;
+  const result = await synthesizeSsmlSafe(
+    {
+      synthesizeSsml: async () => {
+        calls += 1;
+        return { audioData: new ArrayBuffer(0), durationMs: 0 };
+      },
+    },
+    "<speak>",
+  );
+
+  assert.equal(result.ok, false);
+  assert.equal(result.status, "validation-error");
+  assert.equal(calls, 0);
+});
+
+test("synthesizeSsmlSafe returns a successful result for valid SSML", async () => {
+  const expected = { audioData: audio([1]), durationMs: 10 };
+  const result = await synthesizeSsmlSafe(
+    { synthesizeSsml: async () => expected },
+    '<speak version="1.0" xml:lang="en-US"><voice name="en-US-JennyNeural">Hello</voice></speak>',
+  );
+
+  assert.equal(result.ok, true);
+  if (result.ok) assert.equal(result.value, expected);
+});