echogarden 2.6.0 → 2.8.0

package/dist/utilities/WikipediaReader.js

@@ -1,4 +1,4 @@
-import { splitToParagraphs, wordCharacterPattern } from '../nlp/Segmentation.js';
+import { splitToParagraphs, wordCharacterRegExp } from '../nlp/Segmentation.js';
 import { Logger } from './Logger.js';
 export async function parseWikipediaArticle(articleName, language) {
     const logger = new Logger();
@@ -12,13 +12,13 @@ export async function parseWikipediaArticle(articleName, language) {
     const sectionsText = [];
     for (const section of sections) {
         const sectionTitle = section.title();
-        if (wordCharacterPattern.test(sectionTitle)) {
+        if (wordCharacterRegExp.test(sectionTitle)) {
             sectionsText.push(sectionTitle);
         }
         const sectionParagraphs = splitToParagraphs(section.text(), 'single', 'preserve');
         for (const paragraph of sectionParagraphs) {
             const paragraphText = paragraph;
-            if (wordCharacterPattern.test(paragraphText)) {
+            if (wordCharacterRegExp.test(paragraphText)) {
                 sectionsText.push(paragraphText);
             }
         }

package/dist/utilities/WikipediaReader.js.map

@@ -1 +1 @@
-{"version":3,"file":"WikipediaReader.js","sourceRoot":"","sources":["../../src/utilities/WikipediaReader.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAChF,OAAO,EAAE,MAAM,EAAE,MAAM,aAAa,CAAA;AAEpC,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,WAAmB,EAAE,QAAgB;IAChF,MAAM,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;IAE3B,MAAM,MAAM,CAAC,UAAU,CAAC,4BAA4B,CAAC,CAAA;IAErD,MAAM,EAAE,OAAO,EAAE,GAAG,EAAE,GAAG,MAAM,MAAM,CAAC,eAAe,CAAC,CAAA;IAEtD,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAA;IAEvD,IAAI,CAAC,QAAQ,EAAE,CAAC;QACf,MAAM,IAAI,KAAK,CAAC,kCAAkC,CAAC,CAAA;IACpD,CAAC;IAED,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,EAAE,CAAA;IACpC,MAAM,YAAY,GAAa,EAAE,CAAA;IAEjC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAChC,MAAM,YAAY,GAAG,OAAO,CAAC,KAAK,EAAE,CAAA;QAEpC,IAAI,oBAAoB,CAAC,IAAI,CAAC,YAAY,CAAC,EAAE,CAAC;YAC7C,YAAY,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;QAChC,CAAC;QAED,MAAM,iBAAiB,GAAG,iBAAiB,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,UAAU,CAAC,CAAA;QAEjF,KAAK,MAAM,SAAS,IAAI,iBAAiB,EAAE,CAAC;YAC3C,MAAM,aAAa,GAAG,SAAS,CAAA;YAE/B,IAAI,oBAAoB,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;gBAC9C,YAAY,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;YACjC,CAAC;QACF,CAAC;IACF,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,YAAY,CAAA;AACpB,CAAC"}
+{"version":3,"file":"WikipediaReader.js","sourceRoot":"","sources":["../../src/utilities/WikipediaReader.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAA;AAC/E,OAAO,EAAE,MAAM,EAAE,MAAM,aAAa,CAAA;AAEpC,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,WAAmB,EAAE,QAAgB;IAChF,MAAM,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;IAE3B,MAAM,MAAM,CAAC,UAAU,CAAC,4BAA4B,CAAC,CAAA;IAErD,MAAM,EAAE,OAAO,EAAE,GAAG,EAAE,GAAG,MAAM,MAAM,CAAC,eAAe,CAAC,CAAA;IAEtD,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAA;IAEvD,IAAI,CAAC,QAAQ,EAAE,CAAC;QACf,MAAM,IAAI,KAAK,CAAC,kCAAkC,CAAC,CAAA;IACpD,CAAC;IAED,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,EAAE,CAAA;IACpC,MAAM,YAAY,GAAa,EAAE,CAAA;IAEjC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAChC,MAAM,YAAY,GAAG,OAAO,CAAC,KAAK,EAAE,CAAA;QAEpC,IAAI,mBAAmB,CAAC,IAAI,CAAC,YAAY,CAAC,EAAE,CAAC;YAC5C,YAAY,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;QAChC,CAAC;QAED,MAAM,iBAAiB,GAAG,iBAAiB,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,UAAU,CAAC,CAAA;QAEjF,KAAK,MAAM,SAAS,IAAI,iBAAiB,EAAE,CAAC;YAC3C,MAAM,aAAa,GAAG,SAAS,CAAA;YAE/B,IAAI,mBAAmB,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;gBAC7C,YAAY,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;YACjC,CAAC;QACF,CAAC;IACF,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,YAAY,CAAA;AACpB,CAAC"}

package/dist/voice-activity-detection/WebRtcVAD.js

@@ -1,5 +1,5 @@
 import * as AudioBufferConversion from '../audio/AudioBufferConversion.js';
-import { WasmMemoryManager } from '../utilities/WasmMemoryManager.js';
+import { wrapEmscriptenModuleHeap } from 'wasm-heap-manager';
 export async function detectVoiceActivity(rawAudio, frameDuration = 10, mode = 0) {
     if (rawAudio.sampleRate != 16000) {
         throw new Error('Audio sample rate must be 16KHz');
@@ -9,7 +9,7 @@ export async function detectVoiceActivity(rawAudio, frameDuration = 10, mode = 0
 }
 export async function fvad(samples, sampleRate, frameDuration, mode) {
     const m = await getFvadInstance();
-    const wasmMemory = new WasmMemoryManager(m);
+    const wasmHeap = wrapEmscriptenModuleHeap(m);
     const fvad_new = m._fvad_new;
     const fvad_free = m._fvad_free;
     const fvad_reset = m._fvad_reset;
@@ -26,7 +26,7 @@ export async function fvad(samples, sampleRate, frameDuration, mode) {
         throw new Error(`fvad_set_mode failed for mode ${mode}`);
     }
     const frameSampleCount = Math.floor(sampleRate * (frameDuration / 1000));
-    const frameSamplesRef = wasmMemory.allocInt16Array(frameSampleCount);
+    const frameSamplesRef = wasmHeap.allocInt16Array(frameSampleCount);
     const result = [];
     for (let sampleOffset = 0; sampleOffset < samples.length; sampleOffset += frameSampleCount) {
         const frame = samples.subarray(sampleOffset, sampleOffset + frameSampleCount);
@@ -39,7 +39,7 @@ export async function fvad(samples, sampleRate, frameDuration, mode) {
         result.push(fvadResult);
     }
     fvad_free(instancePtr);
-    wasmMemory.freeAll();
+    wasmHeap.freeAll();
     return result;
 }
 let fvadInstance;

package/dist/voice-activity-detection/WebRtcVAD.js.map

@@ -1 +1 @@
-{"version":3,"file":"WebRtcVAD.js","sourceRoot":"","sources":["../../src/voice-activity-detection/WebRtcVAD.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,qBAAqB,MAAM,mCAAmC,CAAA;AAE1E,OAAO,EAAE,iBAAiB,EAAE,MAAM,mCAAmC,CAAA;AAErE,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,QAAkB,EAAE,gBAA8B,EAAE,EAAE,OAAsB,CAAC;IACtH,IAAI,QAAQ,CAAC,UAAU,IAAI,KAAK,EAAE,CAAC;QAClC,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAA;IACnD,CAAC;IAED,MAAM,oBAAoB,GAAc,MAAM,IAAI,CAAC,qBAAqB,CAAC,iBAAiB,CAAC,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,CAAC,UAAU,EAAE,aAAa,EAAE,IAAI,CAAC,CAAA;IAEhK,OAAO,oBAAoB,CAAA;AAC5B,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,IAAI,CAAC,OAAmB,EAAE,UAAkB,EAAE,aAA2B,EAAE,IAAmB;IACnH,MAAM,CAAC,GAAG,MAAM,eAAe,EAAE,CAAA;IAEjC,MAAM,UAAU,GAAG,IAAI,iBAAiB,CAAC,CAAC,CAAC,CAAA;IAE3C,MAAM,QAAQ,GAAG,CAAC,CAAC,SAAS,CAAA;IAC5B,MAAM,SAAS,GAAG,CAAC,CAAC,UAAU,CAAA;IAC9B,MAAM,UAAU,GAAG,CAAC,CAAC,WAAW,CAAA;IAChC,MAAM,aAAa,GAAG,CAAC,CAAC,cAAc,CAAA;IACtC,MAAM,oBAAoB,GAAG,CAAC,CAAC,qBAAqB,CAAA;IACpD,MAAM,YAAY,GAAG,CAAC,CAAC,aAAa,CAAA;IAEpC,MAAM,WAAW,GAAG,QAAQ,EAAE,CAAA;IAE9B,MAAM,eAAe,GAAG,oBAAoB,CAAC,WAAW,EAAE,UAAU,CAAC,CAAA;IAErE,IAAI,eAAe,IAAI,CAAC,EAAE,CAAC;QAC1B,MAAM,IAAI,KAAK,CAAC,yCAAyC,UAAU,EAAE,CAAC,CAAA;IACvE,CAAC;IAED,MAAM,SAAS,GAAG,aAAa,CAAC,WAAW,EAAE,IAAI,CAAC,CAAA;IAElD,IAAI,SAAS,IAAI,CAAC,EAAE,CAAC;QACpB,MAAM,IAAI,KAAK,CAAC,iCAAiC,IAAI,EAAE,CAAC,CAAA;IACzD,CAAC;IAED,MAAM,gBAAgB,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,aAAa,GAAG,IAAI,CAAC,CAAC,CAAA;IACxE,MAAM,eAAe,GAAG,UAAU,CAAC,eAAe,CAAC,gBAAgB,CAAC,CAAA;IAEpE,MAAM,MAAM,GAAG,EAAE,CAAA;IAEjB,KAAK,IAAI,YAAY,GAAG,CAAC,EAAE,YAAY,GAAG,OAAO,CAAC,MAAM,EAAE,YAAY,IAAI,gBAAgB,EAAE,CAAC;QAC5F,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,YAAY,EAAE,YAAY,GAAG,gBAAgB,CAAC,CAAA;QAE7E,eAAe,CAAC,KAAK,EAAE,CAAA;QACvB,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;QAE/B,MAAM,UAAU,GAAG,YAAY,CAAC,WAAW,EAAE,eAAe,CAAC,OAAO,EAAE,gBAAgB,CAAC,CAAA;QAEvF,IAAI,UAAU,IAAI,CAAC,CAAC,EAAE,CAAC;YACtB,MAAM,IAAI,KAAK,CAAC,qBAAqB,CAAC,CAAA;QACvC,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;IACxB,CAAC;IAED,SAAS,CAAC,WAAW,CAAC,CAAA;IACtB,UAAU,CAAC,OAAO,EAAE,CAAA;IAEpB,OAAO,MAAM,CAAA;AACd,CAAC;AAED,IAAI,YAAiB,CAAA;AACrB,KAAK,UAAU,eAAe;IAC7B,IAAI,CAAC,YAAY,EAAE,CAAC;QACnB,MAAM,EAAE,OAAO,EAAE,eAAe,EAAE,GAAG,MAAM,MAAM,CAAC,uBAAuB,CAAC,CAAA;QAE1E,YAAY,GAAG,MAAM,eAAe,EAAE,CAAA;IACvC,CAAC;IAED,OAAO,YAAY,CAAA;AACpB,CAAC"}
+{"version":3,"file":"WebRtcVAD.js","sourceRoot":"","sources":["../../src/voice-activity-detection/WebRtcVAD.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,qBAAqB,MAAM,mCAAmC,CAAA;AAG1E,OAAO,EAAE,wBAAwB,EAAE,MAAM,mBAAmB,CAAA;AAE5D,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,QAAkB,EAAE,gBAA8B,EAAE,EAAE,OAAsB,CAAC;IACtH,IAAI,QAAQ,CAAC,UAAU,IAAI,KAAK,EAAE,CAAC;QAClC,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAA;IACnD,CAAC;IAED,MAAM,oBAAoB,GAAc,MAAM,IAAI,CAAC,qBAAqB,CAAC,iBAAiB,CAAC,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,CAAC,UAAU,EAAE,aAAa,EAAE,IAAI,CAAC,CAAA;IAEhK,OAAO,oBAAoB,CAAA;AAC5B,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,IAAI,CAAC,OAAmB,EAAE,UAAkB,EAAE,aAA2B,EAAE,IAAmB;IACnH,MAAM,CAAC,GAAG,MAAM,eAAe,EAAE,CAAA;IAEjC,MAAM,QAAQ,GAAG,wBAAwB,CAAC,CAAC,CAAC,CAAA;IAE5C,MAAM,QAAQ,GAAG,CAAC,CAAC,SAAS,CAAA;IAC5B,MAAM,SAAS,GAAG,CAAC,CAAC,UAAU,CAAA;IAC9B,MAAM,UAAU,GAAG,CAAC,CAAC,WAAW,CAAA;IAChC,MAAM,aAAa,GAAG,CAAC,CAAC,cAAc,CAAA;IACtC,MAAM,oBAAoB,GAAG,CAAC,CAAC,qBAAqB,CAAA;IACpD,MAAM,YAAY,GAAG,CAAC,CAAC,aAAa,CAAA;IAEpC,MAAM,WAAW,GAAG,QAAQ,EAAE,CAAA;IAE9B,MAAM,eAAe,GAAG,oBAAoB,CAAC,WAAW,EAAE,UAAU,CAAC,CAAA;IAErE,IAAI,eAAe,IAAI,CAAC,EAAE,CAAC;QAC1B,MAAM,IAAI,KAAK,CAAC,yCAAyC,UAAU,EAAE,CAAC,CAAA;IACvE,CAAC;IAED,MAAM,SAAS,GAAG,aAAa,CAAC,WAAW,EAAE,IAAI,CAAC,CAAA;IAElD,IAAI,SAAS,IAAI,CAAC,EAAE,CAAC;QACpB,MAAM,IAAI,KAAK,CAAC,iCAAiC,IAAI,EAAE,CAAC,CAAA;IACzD,CAAC;IAED,MAAM,gBAAgB,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,GAAG,CAAC,aAAa,GAAG,IAAI,CAAC,CAAC,CAAA;IACxE,MAAM,eAAe,GAAG,QAAQ,CAAC,eAAe,CAAC,gBAAgB,CAAC,CAAA;IAElE,MAAM,MAAM,GAAG,EAAE,CAAA;IAEjB,KAAK,IAAI,YAAY,GAAG,CAAC,EAAE,YAAY,GAAG,OAAO,CAAC,MAAM,EAAE,YAAY,IAAI,gBAAgB,EAAE,CAAC;QAC5F,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,YAAY,EAAE,YAAY,GAAG,gBAAgB,CAAC,CAAA;QAE7E,eAAe,CAAC,KAAK,EAAE,CAAA;QACvB,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;QAE/B,MAAM,UAAU,GAAG,YAAY,CAAC,WAAW,EAAE,eAAe,CAAC,OAAO,EAAE,gBAAgB,CAAC,CAAA;QAEvF,IAAI,UAAU,IAAI,CAAC,CAAC,EAAE,CAAC;YACtB,MAAM,IAAI,KAAK,CAAC,qBAAqB,CAAC,CAAA;QACvC,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;IACxB,CAAC;IAED,SAAS,CAAC,WAAW,CAAC,CAAA;IACtB,QAAQ,CAAC,OAAO,EAAE,CAAA;IAElB,OAAO,MAAM,CAAA;AACd,CAAC;AAED,IAAI,YAAiB,CAAA;AACrB,KAAK,UAAU,eAAe;IAC7B,IAAI,CAAC,YAAY,EAAE,CAAC;QACnB,MAAM,EAAE,OAAO,EAAE,eAAe,EAAE,GAAG,MAAM,MAAM,CAAC,uBAAuB,CAAC,CAAA;QAE1E,YAAY,GAAG,MAAM,eAAe,EAAE,CAAA;IACvC,CAAC;IAED,OAAO,YAAY,CAAA;AACpB,CAAC"}

package/docs/Tasklist.md

@@ -7,7 +7,6 @@
 * IPA -> Kirshenbaum translation is still not completely similar to what is output by eSpeak. Also, in rare situations, it outputs characters that are not accepted by eSpeak and eSpeak errors. Investigate when that happens and how to improve on this
 
 ### Subtitles
-* Phrase splitting may split on number separators like the `,` in `100,000`. The new segmentation library would resolve that
 
 ### Browser extension
 * Investigate why WebSpeech events sometimes completely stop working in the middle of an utterance for no apparent reason. Sometimes this is permanent, until the extension is restarted. Is this a browser issue?
@@ -78,7 +77,6 @@
 
 ### API
 * Validate timelines to ensure timestamps are always increasing: no negative timestamps or timestamps over the duration of the audio. No sentences without words, etc. and correct if needed
-* See whether it's possible to detect and include / remove Emoji characters in timelines
 * Add support for phrases in timelines
 * Accept voice list caching options in `SynthesisOptions`
 
@@ -93,7 +91,6 @@
 
 ### Subtitles
 * Split long words if needed
-* Clauses shouldn't be split in the middle of numbers, like the `,` in `123,456`
 * Decide how many punctuation characters to allow before breaking to a new line (currently it's infinite)
 * If a subtitle is too short and at the end of the audio, try to extend it back if possible (for example, if the previous subtitle is already extended, take back from it)
 * Add more clause separators, for even more special cases
@@ -104,7 +101,6 @@
 * Option to disable alignment (only for some engines). Alternative: use a low granularity DTW setting that is very fast to compute
 * Find places to add commas (",") to improve speech fluency. VITS voices don't normally add speech breaks if there is no punctuation
 * An isolated dash " - " can be converted to a " , " to ensure there's a break in the speech
-* Ensure abbreviations like "Ph.d" or similar names are segmented and read correctly (does `cldr` treat it as a word? Maybe eSpeak doesn't recognize it as a word). "C#" and ".NET" as well
 * Find a way to manually reset voice list cache
 * When synthesized text isn't pre-split to sentences, apply sentence splits by using the existing method to convert the output of word timelines to sentence/segment timelines
 * Some `sapi` voices and `msspeech` languages output phones that are converted to Microsoft alphabet, not IPA symbols. Try to see if these can be translated to IPA
@@ -120,7 +116,7 @@
 ### Synthesis / preprocessing
 * Full date normalization (e.g. `21 August 2023`, `21 Aug 2023`, `August 21, 2023`)
 * Add support for capitalized-only rules, and possibly also all uppercase / all lowercase rules
-* Add support for multiple words in `precededBy` and `succeededBy`
+* Add support for multiple consecutive words in `precededBy` and `followedBy` conditions
 * Support substituting to graphemes in lexicons, not only phonemes
 * Cache lexicons to avoid parsing the JSON each time it is loaded (this may not be needed for if the file is relatively small)
 * Is it possible to pre-phonemize common words like "the" or is it a bad idea / not necessary?
@@ -159,7 +155,6 @@
 
 ### Source separation / MDX-NET
 * Option to customize overlap
-* Add more models
 
 ### Server
 * Option to allow or disallow local file paths as arguments to API methods (as a security safeguard)

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "echogarden",
-	"version": "2.6.0",
+	"version": "2.8.0",
 	"description": "An easy-to-use speech toolset. Includes tools for synthesis, recognition, alignment, speech translation, language detection, source separation and more.",
 	"author": "Rotem Dan",
 	"license": "GPL-3.0",
@@ -52,27 +52,27 @@
 		"echogarden": "./dist/cli/CLILauncher.js"
 	},
 	"dependencies": {
-		"@aws-sdk/client-polly": "^3.799.0",
-		"@aws-sdk/client-transcribe-streaming": "^3.799.0",
-		"@echogarden/audio-io": "^0.3.0",
+		"@aws-sdk/client-polly": "^3.810.0",
+		"@aws-sdk/client-transcribe-streaming": "^3.810.0",
+		"@echogarden/audio-io": "^0.4.1",
 		"@echogarden/espeak-ng-emscripten": "^0.3.3",
 		"@echogarden/fasttext-wasm": "^0.1.0",
 		"@echogarden/flite-wasi": "^0.1.1",
 		"@echogarden/fvad-wasm": "^0.2.0",
 		"@echogarden/gnuspeech-wasm": "^0.2.1",
+		"@echogarden/icu-segmentation-wasm": "^0.2.1",
 		"@echogarden/pffft-wasm": "^0.4.2",
 		"@echogarden/rnnoise-wasm": "^0.2.0",
 		"@echogarden/rubberband-wasm": "^0.2.0",
 		"@echogarden/sonic-wasm": "^0.2.0",
 		"@echogarden/speex-resampler-wasm": "^0.3.0",
 		"@echogarden/svoxpico-wasm": "^0.2.0",
-		"@echogarden/transformers-nodejs-lite": "^2.17.1-lite.3",
+		"@echogarden/text-segmentation": "^0.3.1",
+		"@echogarden/transformers-nodejs-lite": "^2.17.1-lite.4",
+		"@echogarden/wave-codec": "^0.3.3",
 		"@mozilla/readability": "^0.6.0",
-		"alawmulaw": "^6.0.0",
 		"chalk": "^5.4.1",
-		"cldr-segmentation": "^2.2.1",
 		"command-exists": "^1.2.9",
-		"compromise": "^14.14.4",
 		"fs-extra": "^11.3.0",
 		"gaxios": "^6.7.1",
 		"graceful-fs": "^4.2.11",
@@ -84,14 +84,16 @@
 		"kuromoji": "^0.1.2",
 		"microsoft-cognitiveservices-speech-sdk": "^1.43.1",
 		"msgpack-lite": "^0.1.26",
-		"onnxruntime-node": "^1.21.1",
-		"openai": "^4.97.0",
+		"onnxruntime-node": "^1.22.0",
+		"openai": "^4.98.0",
+		"regexp-composer": "^0.3.0",
 		"sam-js": "^0.3.1",
 		"strip-ansi": "^7.1.0",
 		"tar": "^7.4.3",
 		"tiktoken": "^1.0.21",
 		"tinyld": "^1.3.4",
 		"wasm-feature-detect": "^1.8.0",
+		"wasm-heap-manager": "^0.2.2",
 		"ws": "^8.18.2",
 		"wtf_wikipedia": "^10.4.0"
 	},
@@ -112,7 +114,7 @@
 		"@types/graceful-fs": "^4.1.9",
 		"@types/jsdom": "^21.1.7",
 		"@types/msgpack-lite": "^0.1.11",
-		"@types/node": "^22.15.3",
+		"@types/node": "^22.15.18",
 		"@types/tar": "^6.1.13",
 		"@types/ws": "^8.18.1",
 		"ts-json-schema-generator": "^2.4.0",

package/src/alignment/SemanticTextAlignment.ts

@@ -3,7 +3,7 @@ import { Logger } from '../utilities/Logger.js'
 import { loadPackage } from '../utilities/PackageManager.js'
 import { alignDTWWindowed } from './DTWSequenceAlignmentWindowed.js'
 import { cosineDistance } from '../math/VectorMath.js'
-import { isPunctuation, isWord, splitToSentences, splitToWords } from '../nlp/Segmentation.js'
+import { includesPunctuation, isWord, parseText } from '../nlp/Segmentation.js'
 import { Timeline, extractEntries } from '../utilities/Timeline.js'
 
 export async function alignTimelineToTextSemantically(timeline: Timeline, text: string, textLangCode: string) {
@@ -26,13 +26,12 @@ export async function alignTimelineToTextSemantically(timeline: Timeline, text:
 
 	const timelineWordEntriesFiltered = timelineWordEntryGroups.flat()
 
-	const textSentences = splitToSentences(text, textLangCode)
+	const segmentedText = await parseText(text, textLangCode)
 
 	const textWordGroups: string[][] = []
 
-	for (const sentenceText of textSentences) {
-		let wordGroup = await splitToWords(sentenceText, textLangCode)
-		wordGroup = wordGroup.filter(word => isWord(word))
+	for (const sentenceEntry of segmentedText.sentences) {
+		const wordGroup = sentenceEntry.words.nonPunctuationWords
 
 		textWordGroups.push(wordGroup)
 	}
@@ -197,8 +196,8 @@ export async function alignWordsToWordsSemantically(wordsGroups1: string[][], wo
 
 	// Align
 	function costFunction(a: TokenEmbeddingData, b: TokenEmbeddingData) {
-		const aIsPunctuation = isPunctuation(a.text)
-		const bIsPunctuation = isPunctuation(b.text)
+		const aIsPunctuation = includesPunctuation(a.text)
+		const bIsPunctuation = includesPunctuation(b.text)
 
 		if (aIsPunctuation === bIsPunctuation) {
 			return cosineDistance(a.embeddingVector, b.embeddingVector)

package/src/api/Alignment.ts

@@ -5,7 +5,7 @@ import { AudioSourceParam, RawAudio, ensureRawAudio, getRawAudioDuration, normal
 import { Logger } from '../utilities/Logger.js'
 
 import * as API from './API.js'
-import { Timeline, addTimeOffsetToTimeline, addWordTextOffsetsToTimeline, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
+import { Timeline, addTimeOffsetToTimeline, addWordTextOffsetsToTimelineInPlace, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
 import { formatLanguageCodeWithName, getDefaultDialectForLanguageCodeIfPossible, getShortLanguageCode, parseLangIdentifier } from '../utilities/Locale.js'
 import { type WhisperAlignmentOptions } from '../recognition/WhisperSTT.js'
 import chalk from 'chalk'
@@ -329,7 +329,7 @@ export async function align(input: AudioSourceParam, transcript: string, options
 	}
 
 	// Add text offsets
-	addWordTextOffsetsToTimeline(mappedTimeline, transcript)
+	addWordTextOffsetsToTimelineInPlace(mappedTimeline, transcript)
 
 	// Make segment timeline
 	const { segmentTimeline } = await wordTimelineToSegmentSentenceTimeline(mappedTimeline, transcript, language, options.plainText?.paragraphBreaks, options.plainText?.whitespace)

package/src/api/Recognition.ts

@@ -5,7 +5,7 @@ import { AudioSourceParam, RawAudio, ensureRawAudio, normalizeAudioLevelInPlace,
 import { Logger } from '../utilities/Logger.js'
 
 import * as API from './API.js'
-import { Timeline, addWordTextOffsetsToTimeline, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
+import { Timeline, addWordTextOffsetsToTimelineInPlace, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
 import { formatLanguageCodeWithName, parseLangIdentifier } from '../utilities/Locale.js'
 import { loadPackage } from '../utilities/PackageManager.js'
 import chalk from 'chalk'
@@ -312,7 +312,7 @@ export async function recognize(input: AudioSourceParam, options: RecognitionOpt
 	}
 
 	// Add text offsets
-	addWordTextOffsetsToTimeline(timeline, transcript)
+	addWordTextOffsetsToTimelineInPlace(timeline, transcript)
 
 	// Make segment timeline
 	const { segmentTimeline } = await wordTimelineToSegmentSentenceTimeline(timeline, transcript, languageCode, 'single', 'preserve')

package/src/api/SourceSeparation.ts

@@ -48,7 +48,7 @@ export async function isolate(input: AudioSourceParam, options: SourceSeparation
 			// Release memory for the converted input audio since it's not needed anymore
 			inputRawAudioAs44100Stereo = undefined
 
-			await logger.startAsync(`Convert isolated audio to back original sample rate (${inputRawAudio.sampleRate} Hz) and channel count (${inputRawAudio.audioChannels.length})`)
+			await logger.startAsync(`Convert isolated audio back to original sample rate (${inputRawAudio.sampleRate} Hz) and channel count (${inputRawAudio.audioChannels.length})`)
 			isolatedRawAudio = await ensureRawAudio(isolatedRawAudio, inputRawAudio.sampleRate, inputRawAudio.audioChannels.length)
 
 			await logger.startAsync(`Subtract from original waveform to extract background audio`)

package/src/api/SpeechTranslation.ts

@@ -4,7 +4,7 @@ import { logToStderr } from '../utilities/Utilities.js'
 import { AudioSourceParam, RawAudio, ensureRawAudio, normalizeAudioLevelInPlace, trimAudioEnd } from '../audio/AudioUtilities.js'
 import { Logger } from '../utilities/Logger.js'
 
-import { Timeline, addWordTextOffsetsToTimeline, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
+import { Timeline, addWordTextOffsetsToTimelineInPlace, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
 import { type WhisperOptions } from '../recognition/WhisperSTT.js'
 import { formatLanguageCodeWithName, getShortLanguageCode, normalizeIdentifierToLanguageCode, parseLangIdentifier } from '../utilities/Locale.js'
 import { EngineMetadata } from './Common.js'
@@ -201,7 +201,7 @@ export async function translateSpeech(input: AudioSourceParam, options: SpeechTr
 	}
 
 	if (wordTimeline) {
-		addWordTextOffsetsToTimeline(wordTimeline, transcript)
+		addWordTextOffsetsToTimelineInPlace(wordTimeline, transcript)
 	}
 
 	if (!segmentTimeline) {

package/src/api/Synthesis.ts

@@ -6,7 +6,7 @@ import { clip, sha256AsHex, stringifyAndFormatJson, logToStderr, yieldToEventLoo
 import { RawAudio, concatAudioSegments, downmixToMono, encodeRawAudioToWave, getSamplePeakDecibels, getEmptyRawAudio, getRawAudioDuration, trimAudioEnd, trimAudioStart, attenuateIfClippingInPlace, normalizeAudioLevelInPlace } from '../audio/AudioUtilities.js'
 import { Logger } from '../utilities/Logger.js'
 
-import { isWordOrSymbolWord, splitToParagraphs, splitToSentences } from '../nlp/Segmentation.js'
+import { isWordOrSymbolWord, parseText, splitToParagraphs } from '../nlp/Segmentation.js'
 import { type RubberbandOptions } from '../dsp/Rubberband.js'
 import { loadLexiconsForLanguage } from '../nlp/Lexicon.js'
 
@@ -116,9 +116,9 @@ async function synthesizeSegments(segments: string[], options: SynthesisOptions,
 	let timeOffset = 0
 
 	for (let segmentIndex = 0; segmentIndex < segments.length; segmentIndex++) {
-		const segmentText = segments[segmentIndex].trim()
+		const segmentText = segments[segmentIndex]
 
-		logger.log(`\n${chalk.magentaBright(`Synthesizing segment ${segmentIndex + 1}/${segments.length}`)}: '${segmentText}'`)
+		logger.log(`\n${chalk.magentaBright(`Synthesizing segment ${segmentIndex + 1}/${segments.length}`)}: '${segmentText.trim()}'`)
 
 		const segmentStartTime = timeOffset
 
@@ -132,9 +132,9 @@ async function synthesizeSegments(segments: string[], options: SynthesisOptions,
 
 		let sentences: string[]
 
-		if ((options.splitToSentences || options.engine == 'vits') && !options.ssml) {
-			sentences = splitToSentences(segmentText, options.language!)
-			sentences = sentences.filter(sentence => sentence.trim() != '')
+		if ((options.splitToSentences || options.engine === 'vits' || options.engine === 'kokoro') && !options.ssml) {
+			const parsedText = await parseText(segmentText, options.language!)
+			sentences = parsedText.sentences.map(sentenceEntry => sentenceEntry.text)
 
 			if (sentences.length == 0) {
 				sentences = ['']
@@ -156,7 +156,7 @@ async function synthesizeSegments(segments: string[], options: SynthesisOptions,
 
 			const sentenceText = sentences[sentenceIndex].trim()
 
-			logger.log(`\n${chalk.magentaBright(`Synthesizing sentence ${sentenceIndex + 1}/${sentences.length}`)}: "${sentenceText}"`)
+			logger.log(`\n${chalk.magentaBright(`Synthesizing sentence ${sentenceIndex + 1}/${sentences.length}`)}: "${sentenceText.trim()}"`)
 
 			const sentenceStartTime = timeOffset
 

package/src/api/TimelineTranslationAlignment.ts

@@ -4,7 +4,7 @@ import { SubtitlesConfig } from '../subtitles/Subtitles.js'
 import { formatLanguageCodeWithName, getShortLanguageCode, parseLangIdentifier } from '../utilities/Locale.js'
 import { Logger } from '../utilities/Logger.js'
 import { extendDeep } from '../utilities/ObjectUtilities.js'
-import { Timeline, addWordTextOffsetsToTimeline, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
+import { Timeline, addWordTextOffsetsToTimelineInPlace, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
 import * as API from './API.js'
 
 export async function alignTimelineTranslation(inputTimeline: Timeline, translatedTranscript: string, options: TimelineTranslationAlignmentOptions): Promise<TimelineTranslationAlignmentResult> {
@@ -95,7 +95,7 @@ export async function alignTimelineTranslation(inputTimeline: Timeline, translat
 
 	logger.start(`Postprocess timeline`)
 
-	addWordTextOffsetsToTimeline(mappedWordTimeline, translatedTranscript)
+	addWordTextOffsetsToTimelineInPlace(mappedWordTimeline, translatedTranscript)
 
 	const { segmentTimeline: mappedTimeline } = await wordTimelineToSegmentSentenceTimeline(mappedWordTimeline, translatedTranscript, targetLanguage)
 

package/src/api/TranslationAlignment.ts

@@ -5,7 +5,7 @@ import { AudioSourceParam, RawAudio, ensureRawAudio, normalizeAudioLevelInPlace,
 import { Logger } from '../utilities/Logger.js'
 
 import * as API from './API.js'
-import { Timeline, addWordTextOffsetsToTimeline, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
+import { Timeline, addWordTextOffsetsToTimelineInPlace, wordTimelineToSegmentSentenceTimeline } from '../utilities/Timeline.js'
 import { formatLanguageCodeWithName, getShortLanguageCode, normalizeIdentifierToLanguageCode, parseLangIdentifier } from '../utilities/Locale.js'
 import { type WhisperAlignmentOptions } from '../recognition/WhisperSTT.js'
 import chalk from 'chalk'
@@ -124,7 +124,7 @@ export async function alignTranslation(input: AudioSourceParam, translatedTransc
 	}
 
 	// Add text offsets
-	addWordTextOffsetsToTimeline(mappedTimeline, translatedTranscript)
+	addWordTextOffsetsToTimelineInPlace(mappedTimeline, translatedTranscript)
 
 	// Make segment timeline
 	const { segmentTimeline } = await wordTimelineToSegmentSentenceTimeline(mappedTimeline, translatedTranscript, sourceLanguage, options.plainText?.paragraphBreaks, options.plainText?.whitespace)