echogarden 0.8.0 → 0.8.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/data/schemas/options.json +40 -46
- package/dist/api/API.d.ts +1 -1
- package/dist/api/API.js +1 -1
- package/dist/api/API.js.map +1 -1
- package/dist/api/Alignment.d.ts +1 -0
- package/dist/api/Alignment.js +4 -0
- package/dist/api/Alignment.js.map +1 -1
- package/dist/api/Common.d.ts +13 -0
- package/dist/api/{Globals.js → Common.js} +1 -1
- package/dist/api/Common.js.map +1 -0
- package/dist/api/Denoising.d.ts +1 -1
- package/dist/api/Recognition.d.ts +2 -6
- package/dist/api/Recognition.js +3 -6
- package/dist/api/Recognition.js.map +1 -1
- package/dist/api/Synthesis.d.ts +2 -6
- package/dist/api/Synthesis.js +1 -1
- package/dist/api/Synthesis.js.map +1 -1
- package/dist/api/Translation.d.ts +4 -7
- package/dist/api/Translation.js +4 -7
- package/dist/api/Translation.js.map +1 -1
- package/dist/api/Vad.d.ts +1 -1
- package/dist/audio/AudioPlayer.d.ts +1 -1
- package/dist/audio/AudioPlayer.js +2 -2
- package/dist/audio/AudioPlayer.js.map +1 -1
- package/dist/cli/CLI.js +21 -16
- package/dist/cli/CLI.js.map +1 -1
- package/dist/denoising/RNNoise.js.map +1 -1
- package/dist/math/VectorMath.d.ts +1 -1
- package/dist/math/VectorMath.js +11 -6
- package/dist/math/VectorMath.js.map +1 -1
- package/dist/nlp/Segmentation.d.ts +1 -2
- package/dist/nlp/Segmentation.js.map +1 -1
- package/dist/recognition/WhisperSTT.d.ts +17 -3
- package/dist/recognition/WhisperSTT.js +44 -16
- package/dist/recognition/WhisperSTT.js.map +1 -1
- package/dist/server/Client.js +2 -2
- package/dist/server/Client.js.map +1 -1
- package/dist/text-language-detection/FastTextLanguageDetection.d.ts +1 -1
- package/dist/text-language-detection/FastTextLanguageDetection.js +1 -1
- package/dist/text-language-detection/FastTextLanguageDetection.js.map +1 -1
- package/dist/utilities/Compression.js.map +1 -1
- package/dist/utilities/FileSystem.js +1 -1
- package/dist/utilities/FileSystem.js.map +1 -1
- package/dist/utilities/PackageManager.js +1 -1
- package/dist/utilities/PackageManager.js.map +1 -1
- package/dist/utilities/TarballMaker.js +1 -1
- package/dist/utilities/TarballMaker.js.map +1 -1
- package/dist/utilities/Timeline.d.ts +2 -1
- package/dist/utilities/Timeline.js +4 -4
- package/dist/utilities/Timeline.js.map +1 -1
- package/docs/Options.md +11 -6
- package/docs/Tasklist.md +4 -2
- package/package.json +2 -2
- package/dist/api/Globals.d.ts +0 -7
- package/dist/api/Globals.js.map +0 -1
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"PackageManager.js","sourceRoot":"","sources":["../../src/utilities/PackageManager.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAA;AAC5B,OAAO,EAAE,yBAAyB,EAAE,MAAM,qBAAqB,CAAA;AAC/D,OAAO,EAAE,aAAa,EAAE,aAAa,EAAE,SAAS,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,iBAAiB,CAAA;AAC7F,OAAO,EAAE,OAAO,EAAE,MAAM,
|
|
1
|
+
{"version":3,"file":"PackageManager.js","sourceRoot":"","sources":["../../src/utilities/PackageManager.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAA;AAC5B,OAAO,EAAE,yBAAyB,EAAE,MAAM,qBAAqB,CAAA;AAC/D,OAAO,EAAE,aAAa,EAAE,aAAa,EAAE,SAAS,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,iBAAiB,CAAA;AAC7F,OAAO,EAAE,OAAO,EAAE,MAAM,kBAAkB,CAAA;AAG1C,MAAM,CAAC,KAAK,UAAU,WAAW,CAAC,WAAmB;IACpD,WAAW,GAAG,qCAAqC,CAAC,WAAW,CAAC,CAAA;IAEhE,MAAM,YAAY,GAAG,MAAM,uBAAuB,EAAE,CAAA;IAEpD,MAAM,WAAW,GAAG,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,WAAW,CAAC,CAAA;IAExD,IAAI,UAAU,CAAC,WAAW,CAAC,EAAE;QAC5B,OAAO,WAAW,CAAA;KAClB;IAED,MAAM,QAAQ,GAAG,aAAa,CAAC,OAAO,CAAC,CAAA;IAEvC,MAAM,OAAO,GAAG,EACf,CAAA;IAED,MAAM,OAAO,GAAkB;QAC9B,GAAG,EAAE,GAAG,cAAc,GAAG,WAAW,SAAS;QAC7C,OAAO;KACP,CAAA;IAED,MAAM,yBAAyB,CAC9B,OAAO,EACP,YAAY,EACZ,QAAQ,EACR,WAAW,CACX,CAAA;IAED,OAAO,WAAW,CAAA;AACnB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,WAAmB;IACtD,WAAW,GAAG,qCAAqC,CAAC,WAAW,CAAC,CAAA;IAEhE,MAAM,YAAY,GAAG,MAAM,uBAAuB,EAAE,CAAA;IAEpD,MAAM,WAAW,GAAG,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,WAAW,CAAC,CAAA;IAExD,MAAM,MAAM,CAAC,WAAW,CAAC,CAAA;AAC1B,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,uBAAuB;IAC5C,MAAM,QAAQ,GAAG,aAAa,CAAC,OAAO,CAAC,CAAA;IAEvC,MAAM,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAA;IAEpD,MAAM,SAAS,CAAC,YAAY,CAAC,CAAA;IAE7B,OAAO,YAAY,CAAA;AACpB,CAAC;AAED,MAAM,UAAU,qCAAqC,CAAC,WAAmB;IACxE,MAAM,UAAU,GAAG,4BAA4B,CAAC,WAAW,CAAC,CAAA;IAE5D,IAAI,UAAU,EAAE;QACf,OAAO,WAAW,CAAA;KAClB;IAED,MAAM,kBAAkB,GAAG,0CAA0C,CAAC,WAAW,CAAC,CAAA;IAElF,OAAO,WAAW,GAAG,GAAG,WAAW,IAAI,kBAAkB,EAAE,CAAA;AAC5D,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,WAAmB;IAC/D,OAAO,WAAW,CAAC,KAAK,CAAC,2DAA2D,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;AAC3F,CAAC;AAED,MAAM,UAAU,0CAA0C,CAAC,sBAA8B;IACxF,OAAO,iCAAiC,CAAC,sBAAsB,CAAC,IAAI,iBAAiB,CAAA;AACtF,CAAC;AAED,MAAM,cAAc,GAAG,qEAAqE,CAAA;AAE5F,MAAM,iBAAiB,GAAG,UAAU,CAAA;AAEpC,MAAM,iCAAiC,GAAsC,EAE5E,CAAA"}
|
|
@@ -2,7 +2,7 @@ import path from "path";
|
|
|
2
2
|
import { createTarball } from "./Compression.js";
|
|
3
3
|
import { ensureDir, existsSync, getAppTempDir, move, readdir, stat } from "./FileSystem.js";
|
|
4
4
|
import { ensureAndGetPackagesDir } from "./PackageManager.js";
|
|
5
|
-
import { appName } from "../api/
|
|
5
|
+
import { appName } from "../api/Common.js";
|
|
6
6
|
import { getRandomHexString } from "./Utilities.js";
|
|
7
7
|
const tarballDir = "../resources/tarballs";
|
|
8
8
|
export async function makeTarballsForInstalledPackages(skipIfExists = false) {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"TarballMaker.js","sourceRoot":"","sources":["../../src/utilities/TarballMaker.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,MAAM,CAAA;AACvB,OAAO,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAA;AAChD,OAAO,EAAE,SAAS,EAAE,UAAU,EAAE,aAAa,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAA;AAC3F,OAAO,EAAE,uBAAuB,EAAE,MAAM,qBAAqB,CAAA;AAC7D,OAAO,EAAE,OAAO,EAAE,MAAM,
|
|
1
|
+
{"version":3,"file":"TarballMaker.js","sourceRoot":"","sources":["../../src/utilities/TarballMaker.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,MAAM,CAAA;AACvB,OAAO,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAA;AAChD,OAAO,EAAE,SAAS,EAAE,UAAU,EAAE,aAAa,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAA;AAC3F,OAAO,EAAE,uBAAuB,EAAE,MAAM,qBAAqB,CAAA;AAC7D,OAAO,EAAE,OAAO,EAAE,MAAM,kBAAkB,CAAA;AAC1C,OAAO,EAAE,kBAAkB,EAAE,MAAM,gBAAgB,CAAA;AAEnD,MAAM,UAAU,GAAG,uBAAuB,CAAA;AAE1C,MAAM,CAAC,KAAK,UAAU,gCAAgC,CAAC,YAAY,GAAG,KAAK;IAC1E,MAAM,WAAW,GAAG,MAAM,uBAAuB,EAAE,CAAA;IACnD,MAAM,WAAW,GAAG,MAAM,OAAO,CAAC,WAAW,CAAC,CAAA;IAE9C,KAAK,MAAM,WAAW,IAAI,WAAW,EAAE;QACtC,IAAI,YAAY,IAAI,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,GAAG,WAAW,SAAS,CAAC,CAAC,EAAE;YAC/E,SAAQ;SACR;QAED,MAAM,WAAW,GAAG,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,WAAW,CAAC,CAAA;QACvD,MAAM,kBAAkB,CAAC,WAAW,EAAE,WAAW,CAAC,CAAA;KAClD;AACF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,SAAiB,EAAE,IAAY;IACvE,MAAM,OAAO,GAAG,aAAa,CAAC,OAAO,CAAC,CAAA;IACtC,MAAM,SAAS,CAAC,OAAO,CAAC,CAAA;IACxB,MAAM,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,kBAAkB,CAAC,EAAE,CAAC,CAAC,CAAA;IAE/D,MAAM,aAAa,CAAC,SAAS,EAAE,YAAY,EAAE,IAAI,CAAC,CAAA;IAElD,MAAM,aAAa,GAAG,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,GAAG,IAAI,SAAS,CAAC,CAAA;IAE7D,MAAM,IAAI,CAAC,YAAY,EAAE,aAAa,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;AAC7D,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAAC,OAAe,EAAE,UAAkB;IAClF,KAAK,MAAM,OAAO,IAAI,MAAM,OAAO,CAAC,OAAO,CAAC,EAAE;QAC7C,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,OAAO,CAAC,CAAA;QAE3C,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,CAAA;QAEpC,IAAI,CAAC,QAAQ,CAAC,WAAW,EAAE,EAAE;YAC5B,SAAQ;SACR;QAED,MAAM,WAAW,GAAG,GAAG,UAAU,IAAI,OAAO,EAAE,CAAA;QAE9C,MAAM,kBAAkB,CAAC,OAAO,EAAE,WAAW,CAAC,CAAA;KAC9C;AACF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,OAAe,EAAE,UAAkB;IACnF,KAAK,MAAM,QAAQ,IAAI,MAAM,OAAO,CAAC,OAAO,CAAC,EAAE;QAC9C,MAAM,wBAAwB,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAA;QAE1D,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAA;QAE7C,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,QAAQ,CAAC,CAAA;QAErC,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,EAAE;YACvB,SAAQ;SACR;QAED,MAAM,WAAW,GAAG,GAAG,UAAU,IAAI,wBAAwB,EAAE,CAAA;QAE/D,MAAM,kBAAkB,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAA;KAC/C;AACF,CAAC"}
|
|
@@ -1,7 +1,8 @@
|
|
|
1
|
+
import { ParagraphBreakType, WhitespaceProcessing } from "../api/Common.js";
|
|
1
2
|
export declare function addTimeOffsetToTimeline(targetTimeline: Timeline, timeOffset: number): Timeline;
|
|
2
3
|
export declare function multiplyTimelineByFactor(targetTimeline: Timeline, factor: number): Timeline;
|
|
3
4
|
export declare function roundTimelineProperties(targetTimeline: Timeline, decimalDigits?: number): Timeline;
|
|
4
|
-
export declare function wordTimelineToSegmentSentenceTimeline(wordTimeline: Timeline, transcript: string, language: string): Promise<{
|
|
5
|
+
export declare function wordTimelineToSegmentSentenceTimeline(wordTimeline: Timeline, transcript: string, language: string, paragraphBreaks?: ParagraphBreakType, whitespace?: WhitespaceProcessing): Promise<{
|
|
5
6
|
segmentTimeline: Timeline;
|
|
6
7
|
}>;
|
|
7
8
|
export type TimelineEntryType = "segment" | "paragraph" | "sentence" | "clause" | "phrase" | "word" | "token" | "letter" | "phone" | "subphone";
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import { isWordOrSymbolWord, splitToParagraphs, splitToSentences } from "../nlp/Segmentation.js";
|
|
2
2
|
import { deepClone } from "./ObjectUtilities.js";
|
|
3
3
|
import { roundToDigits } from "./Utilities.js";
|
|
4
4
|
export function addTimeOffsetToTimeline(targetTimeline, timeOffset) {
|
|
@@ -44,8 +44,8 @@ export function roundTimelineProperties(targetTimeline, decimalDigits = 2) {
|
|
|
44
44
|
}
|
|
45
45
|
return roundedTimeline;
|
|
46
46
|
}
|
|
47
|
-
export async function wordTimelineToSegmentSentenceTimeline(wordTimeline, transcript, language) {
|
|
48
|
-
const paragraphs = await splitToParagraphs(transcript,
|
|
47
|
+
export async function wordTimelineToSegmentSentenceTimeline(wordTimeline, transcript, language, paragraphBreaks = 'double', whitespace = 'collapse') {
|
|
48
|
+
const paragraphs = await splitToParagraphs(transcript, paragraphBreaks, whitespace);
|
|
49
49
|
const segments = paragraphs
|
|
50
50
|
.map(segment => splitToSentences(segment, language).map(sentence => sentence.trim()));
|
|
51
51
|
let text = "";
|
|
@@ -80,7 +80,7 @@ export async function wordTimelineToSegmentSentenceTimeline(wordTimeline, transc
|
|
|
80
80
|
for (let wordIndex = 0; wordIndex < wordTimeline.length; wordIndex++) {
|
|
81
81
|
const wordEntry = wordTimeline[wordIndex];
|
|
82
82
|
const wordText = wordEntry.text;
|
|
83
|
-
if (!
|
|
83
|
+
if (!isWordOrSymbolWord(wordText)) {
|
|
84
84
|
continue;
|
|
85
85
|
}
|
|
86
86
|
const indexOfWordInText = text.indexOf(wordText, wordSearchStartOffset);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"Timeline.js","sourceRoot":"","sources":["../../src/utilities/Timeline.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"Timeline.js","sourceRoot":"","sources":["../../src/utilities/Timeline.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAA;AAChG,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAA;AAChD,OAAO,EAAE,aAAa,EAAE,MAAM,gBAAgB,CAAA;AAE9C,MAAM,UAAU,uBAAuB,CAAC,cAAwB,EAAE,UAAkB;IACnF,IAAI,CAAC,cAAc,EAAE;QACpB,OAAO,cAAc,CAAA;KACrB;IAED,MAAM,WAAW,GAAG,SAAS,CAAC,cAAc,CAAC,CAAA;IAE7C,KAAK,MAAM,oBAAoB,IAAI,WAAW,EAAE;QAC/C,oBAAoB,CAAC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,oBAAoB,CAAC,SAAS,GAAG,UAAU,EAAE,CAAC,CAAC,CAAA;QACzF,oBAAoB,CAAC,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,oBAAoB,CAAC,OAAO,GAAG,UAAU,EAAE,CAAC,CAAC,CAAA;QAErF,IAAI,oBAAoB,CAAC,QAAQ,EAAE;YAClC,oBAAoB,CAAC,QAAQ,GAAG,uBAAuB,CAAC,oBAAoB,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAA;SAClG;KACD;IAED,OAAO,WAAW,CAAA;AACnB,CAAC;AAGD,MAAM,UAAU,wBAAwB,CAAC,cAAwB,EAAE,MAAc;IAChF,MAAM,WAAW,GAAG,SAAS,CAAC,cAAc,CAAC,CAAA;IAE7C,KAAK,MAAM,oBAAoB,IAAI,WAAW,EAAE;QAC/C,oBAAoB,CAAC,SAAS,GAAG,oBAAoB,CAAC,SAAS,GAAG,MAAM,CAAA;QACxE,oBAAoB,CAAC,OAAO,GAAG,oBAAoB,CAAC,OAAO,GAAG,MAAM,CAAA;QAEpE,IAAI,oBAAoB,CAAC,QAAQ,EAAE;YAClC,oBAAoB,CAAC,QAAQ,GAAG,wBAAwB,CAAC,oBAAoB,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;SAC/F;KACD;IAED,OAAO,WAAW,CAAA;AACnB,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,cAAwB,EAAE,aAAa,GAAG,CAAC;IAClF,MAAM,eAAe,GAAG,SAAS,CAAC,cAAc,CAAC,CAAA;IAEjD,KAAK,MAAM,KAAK,IAAI,eAAe,EAAE;QACpC,IAAI,KAAK,CAAC,SAAS,EAAE;YACpB,KAAK,CAAC,SAAS,GAAG,aAAa,CAAC,KAAK,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;SAC/D;QAED,IAAI,KAAK,CAAC,OAAO,EAAE;YAClB,KAAK,CAAC,OAAO,GAAG,aAAa,CAAC,KAAK,CAAC,OAAO,EAAE,aAAa,CAAC,CAAA;SAC3D;QAED,IAAI,KAAK,CAAC,UAAU,EAAE;YACrB,KAAK,CAAC,UAAU,GAAG,aAAa,CAAC,KAAK,CAAC,UAAU,EAAE,aAAa,CAAC,CAAA;SACjE;QAED,IAAI,KAAK,CAAC,QAAQ,EAAE;YACnB,KAAK,CAAC,QAAQ,GAAG,uBAAuB,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAA;SACxD;KACD;IAED,OAAO,eAAe,CAAA;AACvB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,qCAAqC,CAAC,YAAsB,EAAE,UAAkB,EAAE,QAAgB,EAAE,kBAAsC,QAAQ,EAAE,aAAmC,UAAU;IACtN,MAAM,UAAU,GAAG,MAAM,iBAAiB,CAAC,UAAU,EAAE,eAAe,EAAE,UAAU,CAAC,CAAA;IAEnF,MAAM,QAAQ,GAAG,UAAU;SACxB,GAAG,CAAC,OAAO,CAAC,EAAE,CACd,gBAAgB,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAClD,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,CAAA;IAErB,IAAI,IAAI,GAAG,EAAE,CAAA;IACb,MAAM,+BAA+B,GAAoB,EAAE,CAAA;IAE3D,MAAM,eAAe,GAAa,EAAE,CAAA;IAEpC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE;QAC/B,MAAM,kBAAkB,GAAa,EAAE,CAAA;QAEvC,MAAM,YAAY,GAAkB;YACnC,IAAI,EAAE,SAAS;YACf,IAAI,EAAE,EAAE;YACR,SAAS,EAAE,CAAC,CAAC;YACb,OAAO,EAAE,CAAC,CAAC;YACX,QAAQ,EAAE,kBAAkB;SAC5B,CAAA;QAED,KAAK,MAAM,QAAQ,IAAI,OAAO,EAAE;YAC/B,MAAM,aAAa,GAAkB;gBACpC,IAAI,EAAE,UAAU;gBAChB,IAAI,EAAE,QAAQ;gBACd,SAAS,EAAE,CAAC,CAAC;gBACb,OAAO,EAAE,CAAC,CAAC;gBACX,QAAQ,EAAE,EAAE;aACZ,CAAA;YAED,KAAK,MAAM,IAAI,IAAI,QAAQ,GAAG,GAAG,EAAE;gBAClC,IAAI,IAAI,IAAI,CAAA;gBACZ,+BAA+B,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;aACnD;YAED,kBAAkB,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;SACtC;QAED,eAAe,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;KAClC;IAED,IAAI,qBAAqB,GAAG,CAAC,CAAA;IAE7B,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,YAAY,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE;QACrE,MAAM,SAAS,GAAG,YAAY,CAAC,SAAS,CAAC,CAAA;QACzC,MAAM,QAAQ,GAAG,SAAS,CAAC,IAAI,CAAA;QAE/B,IAAI,CAAC,kBAAkB,CAAC,QAAQ,CAAC,EAAE;YAClC,SAAQ;SACR;QAED,MAAM,iBAAiB,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,EAAE,qBAAqB,CAAC,CAAA;QAEvE,IAAI,iBAAiB,IAAI,CAAC,CAAC,EAAE;YAC5B,MAAM,IAAI,KAAK,CAAC,2BAA2B,QAAQ,mCAAmC,qBAAqB,EAAE,CAAC,CAAA;SAC9G;QAED,MAAM,mBAAmB,GAAG,+BAA+B,CAAC,iBAAiB,CAAC,CAAA;QAC9E,mBAAmB,CAAC,QAAS,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;QAE7C,qBAAqB,GAAG,iBAAiB,GAAG,QAAQ,CAAC,MAAM,CAAA;KAC3D;IAED,KAAK,MAAM,YAAY,IAAI,eAAe,EAAE;QAC3C,MAAM,gBAAgB,GAAG,YAAY,CAAC,QAAS,CAAA;QAE/C,IAAI,gBAAgB,CAAC,MAAM,IAAI,CAAC,EAAE;YACjC,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAA;SAClD;QAED,KAAK,MAAM,aAAa,IAAI,gBAAgB,EAAE;YAC7C,MAAM,YAAY,GAAG,aAAa,CAAC,QAAS,CAAA;YAE5C,IAAI,YAAY,CAAC,MAAM,IAAI,CAAC,EAAE;gBAC7B,MAAM,IAAI,KAAK,CAAC,8BAA8B,CAAC,CAAA;aAC/C;YAED,aAAa,CAAC,SAAS,GAAG,YAAY,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;YACnD,aAAa,CAAC,OAAO,GAAG,YAAY,CAAC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,OAAO,CAAA;SACrE;QAED,YAAY,CAAC,IAAI,GAAG,gBAAgB,CAAC,GAAG,CAAC,aAAa,CAAC,EAAE,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;QAEvF,YAAY,CAAC,SAAS,GAAG,gBAAgB,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;QACtD,YAAY,CAAC,OAAO,GAAG,gBAAgB,CAAC,gBAAgB,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,OAAO,CAAA;KAC5E;IAED,OAAO,EAAE,eAAe,EAAE,CAAA;AAC3B,CAAC"}
|
package/docs/Options.md
CHANGED
|
@@ -106,8 +106,10 @@ Applicable to CLI command: `transcribe`.
|
|
|
106
106
|
|
|
107
107
|
**Whisper**:
|
|
108
108
|
* `whisper.model`: selects which Whisper model to use. Can be `tiny`, `tiny.en`, `base`, `base.en`, `small`, `small.en`, `medium`, `medium.en`, `large` (same as `large-v2`), `large-v1`, `large-v2`. Defaults to `tiny`
|
|
109
|
-
* `whisper.temperature`: temperature setting for the text decoder. Impacts amount of randomization for token selection. It is recommended to leave at `0.
|
|
110
|
-
* `whisper.prompt`: initial text to give the Whisper model
|
|
109
|
+
* `whisper.temperature`: temperature setting for the text decoder. Impacts amount of randomization for token selection. It is recommended to leave at `0.1` (close to no randomization - almost always chooses the top ranked token) or choose a relatively low value (about `0.25` or lower) for best results. Defaults to `0.1`
|
|
110
|
+
* `whisper.prompt`: initial text to give the Whisper model. Can be a vocabulary, or example of some sort. Note that if the prompt is very similar to the transcript, the model may intentionally avoid producing the transcript tokens as it may assume that they have already been transcribed. Optional
|
|
111
|
+
* `whisper.topCandidateCount`: the number of top candidate tokens to consider. Defaults to `5`
|
|
112
|
+
* `punctuationThreshold`: the minimal probability for a punctuation token, included in the top candidates, to be chosen unconditionally. A lower threshold encourages the model to output more punctuation symbols. Defaults to `0.2`
|
|
111
113
|
|
|
112
114
|
**Vosk**:
|
|
113
115
|
* `vosk.modelPath`: path to the Vosk model to be used
|
|
@@ -140,8 +142,12 @@ Applicable to CLI command: `align`.
|
|
|
140
142
|
* `language`: language code for the audio and transcript ([ISO 639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes)), like `en`, `fr`, `en-US`, `pt-BR`. Auto-detected from transcript if not set
|
|
141
143
|
* `customLexiconPaths`: an array of custom lexicon file paths. Optional
|
|
142
144
|
|
|
145
|
+
**Plain text processing**:
|
|
146
|
+
* `plainText.paragraphBreaks`: split transcript to paragraphs based on single (`single`), or double (`double`) line breaks. Defaults to `double`
|
|
147
|
+
* `plainText.whitespace`: determines how to process whitespace within transcript paragraphs. Can be `preserve` (leave as is), `removeLineBreaks` (convert line breaks to spaces) or `collapse` (convert runs of whitespace characters, including line breaks, to a single space character). Defaults to `collapse`
|
|
148
|
+
|
|
143
149
|
**DTW**:
|
|
144
|
-
* `dtw.windowDuration`: time duration (in seconds) of the Sakoe-Chiba window when performing DTW alignment.
|
|
150
|
+
* `dtw.windowDuration`: time duration (in seconds) of the Sakoe-Chiba window when performing DTW alignment. If the window duration is shorter than about 25% of the audio duration, consider increasing this value, for better results. Note that a higher value would consume quadratically larger amounts of memory. A value of `600` (10 minutes) would already require several Gigabytes of memory when the audio duration is 10 minutes or greater. The estimated memory requirement is shown in the log messages before alignment starts. Defaults to `120`
|
|
145
151
|
|
|
146
152
|
**DTW-RA only**:
|
|
147
153
|
* `dtw.recognition`: prefix for providing custom recognition options when using `dtw-ra` method, for example: setting `dtw.recognition.engine = silero`
|
|
@@ -158,9 +164,8 @@ Applicable to CLI command: `translate-speech`.
|
|
|
158
164
|
* `languageDetection`: prefix to provide custom options for language detection. Options detailed in section for speech language detection
|
|
159
165
|
|
|
160
166
|
**Whisper**:
|
|
161
|
-
|
|
162
|
-
* `whisper
|
|
163
|
-
* `whisper.prompt` initial text to give the Whisper model. See recognition options for more details. Optional
|
|
167
|
+
|
|
168
|
+
* `whisper`: prefix for options for the Whisper model. Same options as detailed in the recognition section above
|
|
164
169
|
|
|
165
170
|
## Language detection
|
|
166
171
|
|
package/docs/Tasklist.md
CHANGED
|
@@ -144,6 +144,7 @@
|
|
|
144
144
|
* Full date normalization (e.g. `21 August 2023`, `21 Aug 2023`)
|
|
145
145
|
* Add support for capitalized-only rules, and possibly also all uppercase / all lowercase rules.
|
|
146
146
|
* Support normalizing to graphemes, not only phonemes
|
|
147
|
+
* Add support for multiple words in `precededBy` and `succeededBy`
|
|
147
148
|
* Cache lexicons to avoid parsing the JSON each time it is loaded (this may not be needed for if the file is relatively small)
|
|
148
149
|
* Is it possible to pre-phonemize common words like "the" or is it a bad idea / not necessary?
|
|
149
150
|
* Add support for text preprocessing for all engines that can benefit from it (possibly including cloud engines).
|
|
@@ -161,7 +162,7 @@
|
|
|
161
162
|
* Currently, when input is set to be SSML, it is wrapped in a `<speak>` tag. Handle the case where the user made their own SSML document wrapped with a `<speak>` tag as well. Currently it may send invalid input to Azure
|
|
162
163
|
|
|
163
164
|
### Recognition
|
|
164
|
-
* Add confidence to each recognized word
|
|
165
|
+
* Add confidence to each recognized word on any engine that supports it
|
|
165
166
|
* Show alternatives when playing in the CLI. Clear current line and rewrite already printed text for alternatives during the speech recognition process
|
|
166
167
|
* Look for good split points using VAD before performing recognition
|
|
167
168
|
* Option to split recognized audio to segments or sentences, as is done with synthesized audio
|
|
@@ -237,9 +238,9 @@
|
|
|
237
238
|
### Recognition
|
|
238
239
|
* Low latency recognition mode. Make the partial transcription available as fast as possible
|
|
239
240
|
* Live input / microphone recognition
|
|
240
|
-
* Live vosk alternatives events
|
|
241
241
|
* Implement beam search for Whisper decoder
|
|
242
242
|
* Implement beam search for Silero decoder
|
|
243
|
+
* Live vosk alternatives events
|
|
243
244
|
* Investigate exporting Whisper models to 16-bit quantized ONNX or a mix of 16-bit and 32-bit
|
|
244
245
|
|
|
245
246
|
### Alignment
|
|
@@ -250,6 +251,7 @@
|
|
|
250
251
|
* Predict timing for individual letters (graphemes) based on phoneme timestamps
|
|
251
252
|
|
|
252
253
|
## Documentation
|
|
254
|
+
* Start documenting the developer API, to import the program as a library. At least as experimental. Point to source code for reference.
|
|
253
255
|
|
|
254
256
|
### CLI
|
|
255
257
|
* Document the `serve` command
|
package/package.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "echogarden",
|
|
3
|
-
"version": "0.8.
|
|
4
|
-
"description": "An integrated speech system, providing a
|
|
3
|
+
"version": "0.8.2",
|
|
4
|
+
"description": "An integrated speech system, providing a set of synthesis, recognition, alignment, and other processing tools, designed to be directly accessible to end-users.",
|
|
5
5
|
"author": "Rotem Dan",
|
|
6
6
|
"license": "GPL-3.0-only",
|
|
7
7
|
"keywords": [
|
package/dist/api/Globals.d.ts
DELETED
package/dist/api/Globals.js.map
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"Globals.js","sourceRoot":"","sources":["../../src/api/Globals.ts"],"names":[],"mappings":"AAAA,MAAM,CAAC,MAAM,OAAO,GAAG,YAAY,CAAA"}
|