@tscaps/engine 0.2.1 → 0.3.0

package/dist/modules/transcription/MediaBunnyAudioDecoder.js

@@ -0,0 +1,76 @@
+import { ALL_FORMATS, AudioSampleSink, BlobSource, Input, } from 'mediabunny';
+import { BatchedMonoResampler } from '../transcription/BatchedMonoResampler';
+/**
+ * Decodes an audio/video Blob to mono `Float32Array` PCM at the
+ * requested sample rate by demuxing the container with mediabunny and
+ * iterating the primary audio track through {@link AudioSampleSink}.
+ * Downmixes and resamples in fixed-size batches via
+ * {@link BatchedMonoResampler} so peak memory is bounded by one batch
+ * of source-rate audio plus the pre-sized target buffer, not by the
+ * whole track — a long video no longer holds ~4× its useful PCM size
+ * in transient buffers on the way to the worker.
+ *
+ * Main-thread only — `OfflineAudioContext` is not available inside
+ * Worker contexts.
+ *
+ * A codec the current browser cannot decode surfaces as a
+ * `DOMException` named `NotSupportedError`, following the WebCodecs
+ * convention. Any other underlying mediabunny / WebCodecs failure is
+ * re-thrown verbatim so callers can inspect its `name` / `cause` to
+ * classify the reason.
+ */
+export class MediaBunnyAudioDecoder {
+    async decode(audio, targetSampleRate, onProgress) {
+        const input = new Input({ formats: ALL_FORMATS, source: new BlobSource(audio) });
+        try {
+            return await this.decodeInput(input, targetSampleRate, onProgress);
+        }
+        finally {
+            input.dispose();
+        }
+    }
+    async decodeInput(input, targetSampleRate, onProgress) {
+        const track = await input.getPrimaryAudioTrack();
+        if (!track)
+            return new Float32Array(0);
+        if (!(await track.canDecode())) {
+            const codec = await track.getCodec();
+            throw new DOMException(`Audio codec is not supported by this browser: ${codec ?? 'unknown'}`, 'NotSupportedError');
+        }
+        const durationSeconds = await track.computeDuration();
+        if (durationSeconds <= 0)
+            return new Float32Array(0);
+        const target = new Float32Array(Math.max(1, Math.ceil(durationSeconds * targetSampleRate)));
+        const resampler = new BatchedMonoResampler(target, targetSampleRate);
+        const written = await this.streamMonoInto(track, resampler, durationSeconds, onProgress);
+        return target.subarray(0, written);
+    }
+    async streamMonoInto(track, resampler, durationSeconds, onProgress) {
+        const sink = new AudioSampleSink(track);
+        for await (const sample of sink.samples(0)) {
+            try {
+                await resampler.append(sample);
+                if (onProgress)
+                    onProgress(this.progressAt(sample, durationSeconds));
+            }
+            finally {
+                sample.close();
+            }
+        }
+        if (onProgress)
+            onProgress(1);
+        return resampler.flush();
+    }
+    progressAt(sample, durationSeconds) {
+        const decodedSeconds = sample.timestamp + sample.numberOfFrames / sample.sampleRate;
+        if (durationSeconds <= 0)
+            return 0;
+        const ratio = decodedSeconds / durationSeconds;
+        if (ratio < 0)
+            return 0;
+        if (ratio > 1)
+            return 1;
+        return ratio;
+    }
+}
+//# sourceMappingURL=MediaBunnyAudioDecoder.js.map

package/dist/modules/transcription/MediaBunnyAudioDecoder.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"MediaBunnyAudioDecoder.js","sourceRoot":"","sources":["../../../src/modules/transcription/MediaBunnyAudioDecoder.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,WAAW,EACX,eAAe,EACf,UAAU,EACV,KAAK,GAGN,MAAM,YAAY,CAAC;AAEpB,OAAO,EAAE,oBAAoB,EAAE,MAAM,6CAA6C,CAAC;AAEnF;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,OAAO,sBAAsB;IACjC,KAAK,CAAC,MAAM,CACV,KAAW,EACX,gBAAwB,EACxB,UAAuC;QAEvC,MAAM,KAAK,GAAG,IAAI,KAAK,CAAC,EAAE,OAAO,EAAE,WAAW,EAAE,MAAM,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;QACjF,IAAI,CAAC;YACH,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,KAAK,EAAE,gBAAgB,EAAE,UAAU,CAAC,CAAC;QACrE,CAAC;gBAAS,CAAC;YACT,KAAK,CAAC,OAAO,EAAE,CAAC;QAClB,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,WAAW,CACvB,KAAY,EACZ,gBAAwB,EACxB,UAAuC;QAEvC,MAAM,KAAK,GAAG,MAAM,KAAK,CAAC,oBAAoB,EAAE,CAAC;QACjD,IAAI,CAAC,KAAK;YAAE,OAAO,IAAI,YAAY,CAAC,CAAC,CAAC,CAAC;QACvC,IAAI,CAAC,CAAC,MAAM,KAAK,CAAC,SAAS,EAAE,CAAC,EAAE,CAAC;YAC/B,MAAM,KAAK,GAAG,MAAM,KAAK,CAAC,QAAQ,EAAE,CAAC;YACrC,MAAM,IAAI,YAAY,CACpB,iDAAiD,KAAK,IAAI,SAAS,EAAE,EACrE,mBAAmB,CACpB,CAAC;QACJ,CAAC;QACD,MAAM,eAAe,GAAG,MAAM,KAAK,CAAC,eAAe,EAAE,CAAC;QACtD,IAAI,eAAe,IAAI,CAAC;YAAE,OAAO,IAAI,YAAY,CAAC,CAAC,CAAC,CAAC;QAErD,MAAM,MAAM,GAAG,IAAI,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,eAAe,GAAG,gBAAgB,CAAC,CAAC,CAAC,CAAC;QAC5F,MAAM,SAAS,GAAG,IAAI,oBAAoB,CAAC,MAAM,EAAE,gBAAgB,CAAC,CAAC;QACrE,MAAM,OAAO,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,SAAS,EAAE,eAAe,EAAE,UAAU,CAAC,CAAC;QACzF,OAAO,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC;IACrC,CAAC;IAEO,KAAK,CAAC,cAAc,CAC1B,KAAsB,EACtB,SAA+B,EAC/B,eAAuB,EACvB,UAAuC;QAEvC,MAAM,IAAI,GAAG,IAAI,eAAe,CAAC,KAAK,CAAC,CAAC;QACxC,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;YAC3C,IAAI,CAAC;gBACH,MAAM,SAAS,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;gBAC/B,IAAI,UAAU;oBAAE,UAAU,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,eAAe,CAAC,CAAC,CAAC;YACvE,CAAC;oBAAS,CAAC;gBACT,MAAM,CAAC,KAAK,EAAE,CAAC;YACjB,CAAC;QACH,CAAC;QACD,IAAI,UAAU;YAAE,UAAU,CAAC,CAAC,CAAC,CAAC;QAC9B,OAAO,SAAS,CAAC,KAAK,EAAE,CAAC;IAC3B,CAAC;IAEO,UAAU,CAAC,MAAmB,EAAE,eAAuB;QAC7D,MAAM,cAAc,GAAG,MAAM,CAAC,SAAS,GAAG,MAAM,CAAC,cAAc,GAAG,MAAM,CAAC,UAAU,CAAC;QACpF,IAAI,eAAe,IAAI,CAAC;YAAE,OAAO,CAAC,CAAC;QACnC,MAAM,KAAK,GAAG,cAAc,GAAG,eAAe,CAAC;QAC/C,IAAI,KAAK,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC;QACxB,IAAI,KAAK,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC;QACxB,OAAO,KAAK,CAAC;IACf,CAAC;CACF"}

package/dist/modules/transcription/ModelFileCache.d.ts

@@ -0,0 +1,20 @@
+/**
+ * Local store for the model files a transcriber downloads, so a later
+ * run reads them back instead of fetching them again.
+ *
+ * Both operations are best-effort by contract: a store that cannot
+ * answer reports a miss, and a store that cannot keep a file resolves
+ * anyway. Neither ever rejects. Losing the copy costs a download on
+ * the next run, which is worth far less than the transcription in
+ * flight.
+ *
+ * Keys are opaque strings chosen by the caller and must round-trip
+ * verbatim between `put` and `match`.
+ */
+export interface ModelFileCache {
+    /** The stored file, or `undefined` when this key was never kept. */
+    match(key: string): Promise<Response | undefined>;
+    /** Keeps `response` under `key`, replacing nothing already there. */
+    put(key: string, response: Response): Promise<void>;
+}
+//# sourceMappingURL=ModelFileCache.d.ts.map

package/dist/modules/transcription/ModelFileCache.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ModelFileCache.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/ModelFileCache.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AACH,MAAM,WAAW,cAAc;IAC7B,oEAAoE;IACpE,KAAK,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,QAAQ,GAAG,SAAS,CAAC,CAAC;IAElD,qEAAqE;IACrE,GAAG,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC;CACrD"}

package/dist/modules/transcription/ModelFileCache.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=ModelFileCache.js.map

package/dist/modules/transcription/ModelFileCache.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ModelFileCache.js","sourceRoot":"","sources":["../../../src/modules/transcription/ModelFileCache.ts"],"names":[],"mappings":""}

package/dist/modules/transcription/ModelFileCacheUnavailableError.d.ts

@@ -0,0 +1,18 @@
+/**
+ * Raised when the runtime exposes no place to keep model files, so
+ * every run has to download them again.
+ *
+ * The Cache Storage API is restricted to secure contexts and withheld
+ * altogether by some privacy modes, so a page served over plain HTTP
+ * from anything other than localhost has nowhere to put them.
+ *
+ * The `.name` string is set explicitly, and exposed as `ERROR_NAME` on
+ * the class, so consumers can recognise the failure across a Worker
+ * boundary (where structured cloning preserves the string fields but
+ * not the class identity) with a single source of truth.
+ */
+export declare class ModelFileCacheUnavailableError extends Error {
+    static readonly ERROR_NAME = "ModelFileCacheUnavailableError";
+    readonly name = "ModelFileCacheUnavailableError";
+}
+//# sourceMappingURL=ModelFileCacheUnavailableError.d.ts.map

package/dist/modules/transcription/ModelFileCacheUnavailableError.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ModelFileCacheUnavailableError.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/ModelFileCacheUnavailableError.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AACH,qBAAa,8BAA+B,SAAQ,KAAK;IACvD,MAAM,CAAC,QAAQ,CAAC,UAAU,oCAAoC;IAC9D,QAAQ,CAAC,IAAI,oCAA6C;CAC3D"}

package/dist/modules/transcription/ModelFileCacheUnavailableError.js

@@ -0,0 +1,18 @@
+/**
+ * Raised when the runtime exposes no place to keep model files, so
+ * every run has to download them again.
+ *
+ * The Cache Storage API is restricted to secure contexts and withheld
+ * altogether by some privacy modes, so a page served over plain HTTP
+ * from anything other than localhost has nowhere to put them.
+ *
+ * The `.name` string is set explicitly, and exposed as `ERROR_NAME` on
+ * the class, so consumers can recognise the failure across a Worker
+ * boundary (where structured cloning preserves the string fields but
+ * not the class identity) with a single source of truth.
+ */
+export class ModelFileCacheUnavailableError extends Error {
+    static ERROR_NAME = 'ModelFileCacheUnavailableError';
+    name = ModelFileCacheUnavailableError.ERROR_NAME;
+}
+//# sourceMappingURL=ModelFileCacheUnavailableError.js.map

package/dist/modules/transcription/ModelFileCacheUnavailableError.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ModelFileCacheUnavailableError.js","sourceRoot":"","sources":["../../../src/modules/transcription/ModelFileCacheUnavailableError.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AACH,MAAM,OAAO,8BAA+B,SAAQ,KAAK;IACvD,MAAM,CAAU,UAAU,GAAG,gCAAgC,CAAC;IACrD,IAAI,GAAG,8BAA8B,CAAC,UAAU,CAAC"}

package/dist/modules/transcription/SrtTranscriber.d.ts

@@ -1,33 +1,42 @@
-import { Document } from '../document/Document';
+import type { Document } from '../document/Document';
 import type { Transcriber, TranscriberOptions, TranscriberProgressEvent } from '../transcription/Transcriber';
 /**
  * Builds a Document by parsing a SubRip (SRT) caption file. Each cue
  * becomes one `Segment` carrying the cue's full time range; the cue's
  * text (multi-line entries joined with a space, formatting tags
- * stripped) becomes one `Line` whose `Word` timings are spread across
- * the cue range proportionally to character length, so each word
- * lights up in sequence rather than all at once.
+ * stripped) becomes one `Line`.
+ *
+ * SubRip times nothing below a cue, so word timings are shared out
+ * across the cue in proportion to how long each word is to write. Each
+ * word lights up in sequence rather than all at once, but where one
+ * begins is a guess, not something the file said.
  *
  * The audio Blob passed to `transcribe` is ignored. Useful when the
  * caption text and timing are already known — burning a hand-authored
  * SRT into a video, replaying captions from a previous run, etc.
  *
- * Throws when a timecode is malformed; returns an empty Document when
- * the source contains no parseable cues.
+ * A block that carries no timecode is skipped and announced through
+ * `onSkippedCueBlock`, so a single broken cue costs its own text and
+ * nothing else. Throws only when the file has blocks and none of them
+ * could be read, or when a timecode itself is malformed.
+ *
+ * Builds its own collaborators rather than receiving them. The
+ * constructor is the surface a consumer of the engine reaches for, and
+ * it asks for a source and nothing else, so this class is where the
+ * parsing side is assembled rather than a participant in someone
+ * else's assembly.
  */
 export declare class SrtTranscriber implements Transcriber {
     private readonly source;
     onProgress?: (event: TranscriberProgressEvent) => void;
+    /**
+     * Fired once per block left out of the document, before
+     * `transcribe` resolves, carrying the block verbatim.
+     */
+    onSkippedCueBlock?: (block: string) => void;
+    private readonly documentBuilder;
+    private readonly wordReader;
     constructor(source: string);
     transcribe(_audio: Blob, _options?: TranscriberOptions): Promise<Document>;
-    private buildDocumentFromSource;
-    private parseCues;
-    private parseCueBlock;
-    private findTimecodeLineIndex;
-    private parseTimecodeLine;
-    private toSeconds;
-    private normalizeCueText;
-    private buildSegmentFromCue;
-    private distributeWordsAcrossCueRange;
 }
 //# sourceMappingURL=SrtTranscriber.d.ts.map

package/dist/modules/transcription/SrtTranscriber.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"SrtTranscriber.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/SrtTranscriber.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAC;AAMtD,OAAO,KAAK,EACV,WAAW,EACX,kBAAkB,EAClB,wBAAwB,EACzB,MAAM,oCAAoC,CAAC;AAa5C;;;;;;;;;;;;;;GAcG;AACH,qBAAa,cAAe,YAAW,WAAW;IAGpC,OAAO,CAAC,QAAQ,CAAC,MAAM;IAFnC,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,wBAAwB,KAAK,IAAI,CAAC;gBAE1B,MAAM,EAAE,MAAM;IAErC,UAAU,CAAC,MAAM,EAAE,IAAI,EAAE,QAAQ,CAAC,EAAE,kBAAkB,GAAG,OAAO,CAAC,QAAQ,CAAC;IAKhF,OAAO,CAAC,uBAAuB;IAM/B,OAAO,CAAC,SAAS;IASjB,OAAO,CAAC,aAAa;IAarB,OAAO,CAAC,qBAAqB;IAI7B,OAAO,CAAC,iBAAiB;IAWzB,OAAO,CAAC,SAAS;IASjB,OAAO,CAAC,gBAAgB;IAIxB,OAAO,CAAC,mBAAmB;IAS3B,OAAO,CAAC,6BAA6B;CAoBtC"}
+{"version":3,"file":"SrtTranscriber.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/SrtTranscriber.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAC;AAM3D,OAAO,KAAK,EACV,WAAW,EACX,kBAAkB,EAClB,wBAAwB,EACzB,MAAM,oCAAoC,CAAC;AAE5C;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,qBAAa,cAAe,YAAW,WAAW;IAepC,OAAO,CAAC,QAAQ,CAAC,MAAM;IAdnC,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,wBAAwB,KAAK,IAAI,CAAC;IAEvD;;;OAGG;IACH,iBAAiB,CAAC,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,IAAI,CAAC;IAE5C,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAmD;IACnF,OAAO,CAAC,QAAQ,CAAC,UAAU,CAGzB;gBAE2B,MAAM,EAAE,MAAM;IAErC,UAAU,CAAC,MAAM,EAAE,IAAI,EAAE,QAAQ,CAAC,EAAE,kBAAkB,GAAG,OAAO,CAAC,QAAQ,CAAC;CAMjF"}

package/dist/modules/transcription/SrtTranscriber.js

@@ -1,111 +1,53 @@
-import { Document } from '../document/Document';
-import { Section } from '../document/Section';
-import { Segment } from '../document/Segment';
-import { Line } from '../document/Line';
-import { Word } from '../document/Word';
-import { TimeFragment } from '../document/TimeFragment';
-const SRT_TIMECODE_RE = /^(\d{1,2}):(\d{2}):(\d{2})[,.](\d{1,3})\s*-->\s*(\d{1,2}):(\d{2}):(\d{2})[,.](\d{1,3})/;
-const FORMATTING_TAG_RE = /<[^>]+>|\{[^}]+\}/g;
-const BOM_RE = /^\uFEFF/;
+import { CueDocumentBuilder } from '../transcription/CueDocumentBuilder';
+import { CueTextTokenizer } from '../transcription/CueTextTokenizer';
+import { CueTimecodeReader } from '../transcription/CueTimecodeReader';
+import { EstimatedCueWordReader } from '../transcription/EstimatedCueWordReader';
+import { WordTimingEstimator } from '../transcription/WordTimingEstimator';
 /**
  * Builds a Document by parsing a SubRip (SRT) caption file. Each cue
  * becomes one `Segment` carrying the cue's full time range; the cue's
  * text (multi-line entries joined with a space, formatting tags
- * stripped) becomes one `Line` whose `Word` timings are spread across
- * the cue range proportionally to character length, so each word
- * lights up in sequence rather than all at once.
+ * stripped) becomes one `Line`.
+ *
+ * SubRip times nothing below a cue, so word timings are shared out
+ * across the cue in proportion to how long each word is to write. Each
+ * word lights up in sequence rather than all at once, but where one
+ * begins is a guess, not something the file said.
  *
  * The audio Blob passed to `transcribe` is ignored. Useful when the
  * caption text and timing are already known — burning a hand-authored
  * SRT into a video, replaying captions from a previous run, etc.
  *
- * Throws when a timecode is malformed; returns an empty Document when
- * the source contains no parseable cues.
+ * A block that carries no timecode is skipped and announced through
+ * `onSkippedCueBlock`, so a single broken cue costs its own text and
+ * nothing else. Throws only when the file has blocks and none of them
+ * could be read, or when a timecode itself is malformed.
+ *
+ * Builds its own collaborators rather than receiving them. The
+ * constructor is the surface a consumer of the engine reaches for, and
+ * it asks for a source and nothing else, so this class is where the
+ * parsing side is assembled rather than a participant in someone
+ * else's assembly.
  */
 export class SrtTranscriber {
     source;
     onProgress;
+    /**
+     * Fired once per block left out of the document, before
+     * `transcribe` resolves, carrying the block verbatim.
+     */
+    onSkippedCueBlock;
+    documentBuilder = new CueDocumentBuilder(new CueTimecodeReader());
+    wordReader = new EstimatedCueWordReader(new CueTextTokenizer(), new WordTimingEstimator());
     constructor(source) {
         this.source = source;
     }
     async transcribe(_audio, _options) {
         this.onProgress?.({ stage: 'inferring', progress: 1 });
-        return this.buildDocumentFromSource(this.source);
-    }
-    buildDocumentFromSource(source) {
-        const segments = this.parseCues(source).map((cue) => this.buildSegmentFromCue(cue));
-        const section = new Section({ segments, kind: '' });
-        return new Document({ sections: [section] });
-    }
-    parseCues(source) {
-        const normalized = source.replace(BOM_RE, '').replace(/\r\n?/g, '\n').trim();
-        if (normalized.length === 0)
-            return [];
-        return normalized
-            .split(/\n{2,}/)
-            .map((block) => this.parseCueBlock(block))
-            .filter((cue) => cue !== null);
-    }
-    parseCueBlock(block) {
-        const lines = block.split('\n').map((l) => l.trim()).filter((l) => l.length > 0);
-        if (lines.length === 0)
-            return null;
-        const timecodeLineIndex = this.findTimecodeLineIndex(lines);
-        if (timecodeLineIndex === -1) {
-            throw new Error(`SRT block has no timecode line: ${JSON.stringify(block)}`);
-        }
-        const range = this.parseTimecodeLine(lines[timecodeLineIndex]);
-        const text = this.normalizeCueText(lines.slice(timecodeLineIndex + 1));
-        if (text.length === 0)
-            return null;
-        return { startSeconds: range.startSeconds, endSeconds: range.endSeconds, text };
-    }
-    findTimecodeLineIndex(lines) {
-        return lines.findIndex((line) => SRT_TIMECODE_RE.test(line));
-    }
-    parseTimecodeLine(line) {
-        const match = SRT_TIMECODE_RE.exec(line);
-        if (match === null)
-            throw new Error(`Malformed SRT timecode: ${JSON.stringify(line)}`);
-        const startSeconds = this.toSeconds(match[1], match[2], match[3], match[4]);
-        const endSeconds = this.toSeconds(match[5], match[6], match[7], match[8]);
-        if (endSeconds < startSeconds) {
-            throw new Error(`SRT timecode ends before it starts: ${JSON.stringify(line)}`);
-        }
-        return { startSeconds, endSeconds };
-    }
-    toSeconds(hours, minutes, seconds, milliseconds) {
-        return (Number(hours) * 3600 +
-            Number(minutes) * 60 +
-            Number(seconds) +
-            Number(milliseconds.padEnd(3, '0')) / 1000);
-    }
-    normalizeCueText(textLines) {
-        return textLines.join(' ').replace(FORMATTING_TAG_RE, '').replace(/\s+/g, ' ').trim();
-    }
-    buildSegmentFromCue(cue) {
-        const tokens = cue.text.split(/\s+/).filter((token) => token.length > 0);
-        if (tokens.length === 0) {
-            return new Segment({ lines: [new Line({ words: [] })] });
-        }
-        const words = this.distributeWordsAcrossCueRange(tokens, cue.startSeconds, cue.endSeconds);
-        return new Segment({ lines: [new Line({ words })] });
-    }
-    distributeWordsAcrossCueRange(tokens, startSeconds, endSeconds) {
-        const totalWeight = tokens.reduce((sum, token) => sum + token.length, 0);
-        const cueDuration = endSeconds - startSeconds;
-        if (totalWeight === 0 || cueDuration <= 0) {
-            return tokens.map((token) => new Word({ text: token, time: new TimeFragment(startSeconds, endSeconds) }));
-        }
-        const words = [];
-        let elapsedWeight = 0;
-        for (const token of tokens) {
-            const wordStart = startSeconds + (elapsedWeight / totalWeight) * cueDuration;
-            elapsedWeight += token.length;
-            const wordEnd = startSeconds + (elapsedWeight / totalWeight) * cueDuration;
-            words.push(new Word({ text: token, time: new TimeFragment(wordStart, wordEnd) }));
-        }
-        return words;
+        const result = this.documentBuilder.build(this.source, this.wordReader);
+        for (const block of result.skippedBlocks)
+            this.onSkippedCueBlock?.(block);
+        return result.document;
     }
 }
 //# sourceMappingURL=SrtTranscriber.js.map

package/dist/modules/transcription/SrtTranscriber.js.map

@@ -1 +1 @@
-{"version":3,"file":"SrtTranscriber.js","sourceRoot":"","sources":["../../../src/modules/transcription/SrtTranscriber.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAC;AACtD,OAAO,EAAE,OAAO,EAAE,MAAM,2BAA2B,CAAC;AACpD,OAAO,EAAE,OAAO,EAAE,MAAM,2BAA2B,CAAC;AACpD,OAAO,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AAC9C,OAAO,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AAC9C,OAAO,EAAE,YAAY,EAAE,MAAM,gCAAgC,CAAC;AAO9D,MAAM,eAAe,GACnB,wFAAwF,CAAC;AAC3F,MAAM,iBAAiB,GAAG,oBAAoB,CAAC;AAC/C,MAAM,MAAM,GAAG,SAAS,CAAC;AAQzB;;;;;;;;;;;;;;GAcG;AACH,MAAM,OAAO,cAAc;IAGI;IAF7B,UAAU,CAA6C;IAEvD,YAA6B,MAAc;QAAd,WAAM,GAAN,MAAM,CAAQ;IAAG,CAAC;IAE/C,KAAK,CAAC,UAAU,CAAC,MAAY,EAAE,QAA6B;QAC1D,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QACvD,OAAO,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACnD,CAAC;IAEO,uBAAuB,CAAC,MAAc;QAC5C,MAAM,QAAQ,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,CAAC,CAAC;QACpF,MAAM,OAAO,GAAG,IAAI,OAAO,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC;QACpD,OAAO,IAAI,QAAQ,CAAC,EAAE,QAAQ,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;IAC/C,CAAC;IAEO,SAAS,CAAC,MAAc;QAC9B,MAAM,UAAU,GAAG,MAAM,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;QAC7E,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QACvC,OAAO,UAAU;aACd,KAAK,CAAC,QAAQ,CAAC;aACf,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC;aACzC,MAAM,CAAC,CAAC,GAAG,EAAoB,EAAE,CAAC,GAAG,KAAK,IAAI,CAAC,CAAC;IACrD,CAAC;IAEO,aAAa,CAAC,KAAa;QACjC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACjF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC;QACpC,MAAM,iBAAiB,GAAG,IAAI,CAAC,qBAAqB,CAAC,KAAK,CAAC,CAAC;QAC5D,IAAI,iBAAiB,KAAK,CAAC,CAAC,EAAE,CAAC;YAC7B,MAAM,IAAI,KAAK,CAAC,mCAAmC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;QAC9E,CAAC;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,iBAAiB,CAAC,KAAK,CAAC,iBAAiB,CAAE,CAAC,CAAC;QAChE,MAAM,IAAI,GAAG,IAAI,CAAC,gBAAgB,CAAC,KAAK,CAAC,KAAK,CAAC,iBAAiB,GAAG,CAAC,CAAC,CAAC,CAAC;QACvE,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC;QACnC,OAAO,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,UAAU,EAAE,KAAK,CAAC,UAAU,EAAE,IAAI,EAAE,CAAC;IAClF,CAAC;IAEO,qBAAqB,CAAC,KAA4B;QACxD,OAAO,KAAK,CAAC,SAAS,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IAC/D,CAAC;IAEO,iBAAiB,CAAC,IAAY;QACpC,MAAM,KAAK,GAAG,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACzC,IAAI,KAAK,KAAK,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,2BAA2B,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACvF,MAAM,YAAY,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,CAAC;QAChF,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,CAAC;QAC9E,IAAI,UAAU,GAAG,YAAY,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CAAC,uCAAuC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACjF,CAAC;QACD,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,CAAC;IACtC,CAAC;IAEO,SAAS,CAAC,KAAa,EAAE,OAAe,EAAE,OAAe,EAAE,YAAoB;QACrF,OAAO,CACL,MAAM,CAAC,KAAK,CAAC,GAAG,IAAI;YACpB,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE;YACpB,MAAM,CAAC,OAAO,CAAC;YACf,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,GAAG,IAAI,CAC3C,CAAC;IACJ,CAAC;IAEO,gBAAgB,CAAC,SAAgC;QACvD,OAAO,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,iBAAiB,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;IACxF,CAAC;IAEO,mBAAmB,CAAC,GAAc;QACxC,MAAM,MAAM,GAAG,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACzE,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACxB,OAAO,IAAI,OAAO,CAAC,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;QAC3D,CAAC;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,6BAA6B,CAAC,MAAM,EAAE,GAAG,CAAC,YAAY,EAAE,GAAG,CAAC,UAAU,CAAC,CAAC;QAC3F,OAAO,IAAI,OAAO,CAAC,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;IACvD,CAAC;IAEO,6BAA6B,CACnC,MAA6B,EAC7B,YAAoB,EACpB,UAAkB;QAElB,MAAM,WAAW,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,EAAE,CAAC,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QACzE,MAAM,WAAW,GAAG,UAAU,GAAG,YAAY,CAAC;QAC9C,IAAI,WAAW,KAAK,CAAC,IAAI,WAAW,IAAI,CAAC,EAAE,CAAC;YAC1C,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,YAAY,CAAC,YAAY,EAAE,UAAU,CAAC,EAAE,CAAC,CAAC,CAAC;QAC5G,CAAC;QACD,MAAM,KAAK,GAAW,EAAE,CAAC;QACzB,IAAI,aAAa,GAAG,CAAC,CAAC;QACtB,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,MAAM,SAAS,GAAG,YAAY,GAAG,CAAC,aAAa,GAAG,WAAW,CAAC,GAAG,WAAW,CAAC;YAC7E,aAAa,IAAI,KAAK,CAAC,MAAM,CAAC;YAC9B,MAAM,OAAO,GAAG,YAAY,GAAG,CAAC,aAAa,GAAG,WAAW,CAAC,GAAG,WAAW,CAAC;YAC3E,KAAK,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,YAAY,CAAC,SAAS,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC;QACpF,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;CACF"}
+{"version":3,"file":"SrtTranscriber.js","sourceRoot":"","sources":["../../../src/modules/transcription/SrtTranscriber.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAE,MAAM,2CAA2C,CAAC;AAC/E,OAAO,EAAE,gBAAgB,EAAE,MAAM,yCAAyC,CAAC;AAC3E,OAAO,EAAE,iBAAiB,EAAE,MAAM,0CAA0C,CAAC;AAC7E,OAAO,EAAE,sBAAsB,EAAE,MAAM,+CAA+C,CAAC;AACvF,OAAO,EAAE,mBAAmB,EAAE,MAAM,4CAA4C,CAAC;AAOjF;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,MAAM,OAAO,cAAc;IAeI;IAd7B,UAAU,CAA6C;IAEvD;;;OAGG;IACH,iBAAiB,CAA2B;IAE3B,eAAe,GAAG,IAAI,kBAAkB,CAAC,IAAI,iBAAiB,EAAE,CAAC,CAAC;IAClE,UAAU,GAAG,IAAI,sBAAsB,CACtD,IAAI,gBAAgB,EAAE,EACtB,IAAI,mBAAmB,EAAE,CAC1B,CAAC;IAEF,YAA6B,MAAc;QAAd,WAAM,GAAN,MAAM,CAAQ;IAAG,CAAC;IAE/C,KAAK,CAAC,UAAU,CAAC,MAAY,EAAE,QAA6B;QAC1D,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QACvD,MAAM,MAAM,GAAG,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC;QACxE,KAAK,MAAM,KAAK,IAAI,MAAM,CAAC,aAAa;YAAE,IAAI,CAAC,iBAAiB,EAAE,CAAC,KAAK,CAAC,CAAC;QAC1E,OAAO,MAAM,CAAC,QAAQ,CAAC;IACzB,CAAC;CACF"}

package/dist/modules/transcription/SubtitleFileUnreadableError.d.ts

@@ -0,0 +1,22 @@
+/**
+ * Raised when a caption file yielded no cues at all and at least one
+ * of its blocks was unreadable.
+ *
+ * Skipping the odd broken block leaves a usable file behind; skipping
+ * every block does not, and handing back an empty transcript would
+ * read as "this file said nothing" rather than "this file could not
+ * be understood".
+ *
+ * A file that is genuinely empty is not this: it yields an empty
+ * document, because there is nothing to misread.
+ *
+ * The `.name` string is set explicitly, and exposed as `ERROR_NAME` on
+ * the class, so consumers can recognise the failure across a Worker
+ * boundary (where structured cloning preserves the string fields but
+ * not the class identity) with a single source of truth.
+ */
+export declare class SubtitleFileUnreadableError extends Error {
+    static readonly ERROR_NAME = "SubtitleFileUnreadableError";
+    readonly name = "SubtitleFileUnreadableError";
+}
+//# sourceMappingURL=SubtitleFileUnreadableError.d.ts.map

package/dist/modules/transcription/SubtitleFileUnreadableError.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"SubtitleFileUnreadableError.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/SubtitleFileUnreadableError.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AACH,qBAAa,2BAA4B,SAAQ,KAAK;IACpD,MAAM,CAAC,QAAQ,CAAC,UAAU,iCAAiC;IAC3D,QAAQ,CAAC,IAAI,iCAA0C;CACxD"}

package/dist/modules/transcription/SubtitleFileUnreadableError.js

@@ -0,0 +1,22 @@
+/**
+ * Raised when a caption file yielded no cues at all and at least one
+ * of its blocks was unreadable.
+ *
+ * Skipping the odd broken block leaves a usable file behind; skipping
+ * every block does not, and handing back an empty transcript would
+ * read as "this file said nothing" rather than "this file could not
+ * be understood".
+ *
+ * A file that is genuinely empty is not this: it yields an empty
+ * document, because there is nothing to misread.
+ *
+ * The `.name` string is set explicitly, and exposed as `ERROR_NAME` on
+ * the class, so consumers can recognise the failure across a Worker
+ * boundary (where structured cloning preserves the string fields but
+ * not the class identity) with a single source of truth.
+ */
+export class SubtitleFileUnreadableError extends Error {
+    static ERROR_NAME = 'SubtitleFileUnreadableError';
+    name = SubtitleFileUnreadableError.ERROR_NAME;
+}
+//# sourceMappingURL=SubtitleFileUnreadableError.js.map

package/dist/modules/transcription/SubtitleFileUnreadableError.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"SubtitleFileUnreadableError.js","sourceRoot":"","sources":["../../../src/modules/transcription/SubtitleFileUnreadableError.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,OAAO,2BAA4B,SAAQ,KAAK;IACpD,MAAM,CAAU,UAAU,GAAG,6BAA6B,CAAC;IAClD,IAAI,GAAG,2BAA2B,CAAC,UAAU,CAAC"}

package/dist/modules/transcription/TimestampedCueWordReader.d.ts

@@ -0,0 +1,27 @@
+import type { Word } from '../document/Word';
+import type { CueTextTokenizer } from '../transcription/CueTextTokenizer';
+import type { CueTimecodeReader } from '../transcription/CueTimecodeReader';
+import type { CueWordReader } from '../transcription/CueWordReader';
+import type { WordTimingEstimator } from '../transcription/WordTimingEstimator';
+/**
+ * Reads a cue from a format that can mark, inside the cue's own text,
+ * when a word is spoken — WebVTT does this so a caption can be revealed
+ * word by word without being broken into an entry per word.
+ *
+ * Each mark cuts the cue into a stretch that runs until the next mark,
+ * and the words of a stretch share it out between them. A mark before
+ * every word therefore yields the timing the file recorded; a mark
+ * every few words yields those, with the words between them estimated.
+ * A cue carrying no marks at all is one stretch, which is the same
+ * answer a format without marks would give.
+ */
+export declare class TimestampedCueWordReader implements CueWordReader {
+    private readonly tokenizer;
+    private readonly timecodeReader;
+    private readonly estimator;
+    constructor(tokenizer: CueTextTokenizer, timecodeReader: CueTimecodeReader, estimator: WordTimingEstimator);
+    read(text: string, startSeconds: number, endSeconds: number): ReadonlyArray<Word>;
+    private cutIntoStretches;
+    private positionOf;
+}
+//# sourceMappingURL=TimestampedCueWordReader.d.ts.map

package/dist/modules/transcription/TimestampedCueWordReader.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"TimestampedCueWordReader.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/TimestampedCueWordReader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AACnD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,yCAAyC,CAAC;AAChF,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,0CAA0C,CAAC;AAClF,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,sCAAsC,CAAC;AAC1E,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,4CAA4C,CAAC;AAYtF;;;;;;;;;;;GAWG;AACH,qBAAa,wBAAyB,YAAW,aAAa;IAE1D,OAAO,CAAC,QAAQ,CAAC,SAAS;IAC1B,OAAO,CAAC,QAAQ,CAAC,cAAc;IAC/B,OAAO,CAAC,QAAQ,CAAC,SAAS;gBAFT,SAAS,EAAE,gBAAgB,EAC3B,cAAc,EAAE,iBAAiB,EACjC,SAAS,EAAE,mBAAmB;IAGjD,IAAI,CAAC,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,aAAa,CAAC,IAAI,CAAC;IAWjF,OAAO,CAAC,gBAAgB;IAuBxB,OAAO,CAAC,UAAU;CAGnB"}

package/dist/modules/transcription/TimestampedCueWordReader.js

@@ -0,0 +1,52 @@
+/** A cue timestamp: a position on its own inside the cue's text. */
+const CUE_TIMESTAMP = /<(?:\d{1,3}:)?\d{2}:\d{2}[.,]\d{1,3}>/g;
+/**
+ * Reads a cue from a format that can mark, inside the cue's own text,
+ * when a word is spoken — WebVTT does this so a caption can be revealed
+ * word by word without being broken into an entry per word.
+ *
+ * Each mark cuts the cue into a stretch that runs until the next mark,
+ * and the words of a stretch share it out between them. A mark before
+ * every word therefore yields the timing the file recorded; a mark
+ * every few words yields those, with the words between them estimated.
+ * A cue carrying no marks at all is one stretch, which is the same
+ * answer a format without marks would give.
+ */
+export class TimestampedCueWordReader {
+    tokenizer;
+    timecodeReader;
+    estimator;
+    constructor(tokenizer, timecodeReader, estimator) {
+        this.tokenizer = tokenizer;
+        this.timecodeReader = timecodeReader;
+        this.estimator = estimator;
+    }
+    read(text, startSeconds, endSeconds) {
+        return this.cutIntoStretches(text, startSeconds, endSeconds)
+            .flatMap((stretch) => [
+            ...this.estimator.spread(this.tokenizer.tokenize(stretch.text), stretch.startSeconds, stretch.endSeconds),
+        ]);
+    }
+    cutIntoStretches(text, startSeconds, endSeconds) {
+        const stretches = [];
+        let textFrom = 0;
+        let stretchStart = startSeconds;
+        for (const mark of text.matchAll(CUE_TIMESTAMP)) {
+            const markAt = this.positionOf(mark[0]);
+            const markFrom = mark.index ?? 0;
+            stretches.push({
+                text: text.slice(textFrom, markFrom),
+                startSeconds: stretchStart,
+                endSeconds: markAt,
+            });
+            textFrom = markFrom + mark[0].length;
+            stretchStart = markAt;
+        }
+        stretches.push({ text: text.slice(textFrom), startSeconds: stretchStart, endSeconds });
+        return stretches;
+    }
+    positionOf(mark) {
+        return this.timecodeReader.readPosition(mark.slice(1, -1));
+    }
+}
+//# sourceMappingURL=TimestampedCueWordReader.js.map

package/dist/modules/transcription/TimestampedCueWordReader.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"TimestampedCueWordReader.js","sourceRoot":"","sources":["../../../src/modules/transcription/TimestampedCueWordReader.ts"],"names":[],"mappings":"AAMA,oEAAoE;AACpE,MAAM,aAAa,GAAG,wCAAwC,CAAC;AAS/D;;;;;;;;;;;GAWG;AACH,MAAM,OAAO,wBAAwB;IAEhB;IACA;IACA;IAHnB,YACmB,SAA2B,EAC3B,cAAiC,EACjC,SAA8B;QAF9B,cAAS,GAAT,SAAS,CAAkB;QAC3B,mBAAc,GAAd,cAAc,CAAmB;QACjC,cAAS,GAAT,SAAS,CAAqB;IAC9C,CAAC;IAEJ,IAAI,CAAC,IAAY,EAAE,YAAoB,EAAE,UAAkB;QACzD,OAAO,IAAI,CAAC,gBAAgB,CAAC,IAAI,EAAE,YAAY,EAAE,UAAU,CAAC;aACzD,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC;YACpB,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CACtB,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,EACrC,OAAO,CAAC,YAAY,EACpB,OAAO,CAAC,UAAU,CACnB;SACF,CAAC,CAAC;IACP,CAAC;IAEO,gBAAgB,CACtB,IAAY,EACZ,YAAoB,EACpB,UAAkB;QAElB,MAAM,SAAS,GAAmB,EAAE,CAAC;QACrC,IAAI,QAAQ,GAAG,CAAC,CAAC;QACjB,IAAI,YAAY,GAAG,YAAY,CAAC;QAChC,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,EAAE,CAAC;YAChD,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YACxC,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,IAAI,CAAC,CAAC;YACjC,SAAS,CAAC,IAAI,CAAC;gBACb,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ,EAAE,QAAQ,CAAC;gBACpC,YAAY,EAAE,YAAY;gBAC1B,UAAU,EAAE,MAAM;aACnB,CAAC,CAAC;YACH,QAAQ,GAAG,QAAQ,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;YACrC,YAAY,GAAG,MAAM,CAAC;QACxB,CAAC;QACD,SAAS,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,YAAY,EAAE,YAAY,EAAE,UAAU,EAAE,CAAC,CAAC;QACvF,OAAO,SAAS,CAAC;IACnB,CAAC;IAEO,UAAU,CAAC,IAAY;QAC7B,OAAO,IAAI,CAAC,cAAc,CAAC,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7D,CAAC;CACF"}

package/dist/modules/transcription/Transcriber.d.ts

@@ -16,8 +16,21 @@ export type TranscriberProgressEvent = {
     stage: 'inferring';
     progress?: number;
 };
+/**
+ * An audio region, in absolute seconds, that transcription deliberately
+ * left empty because decoding produced no trustworthy output for it.
+ */
+export interface UntranscribedRegion {
+    startSeconds: number;
+    endSeconds: number;
+}
 export interface Transcriber {
     onProgress?: (event: TranscriberProgressEvent) => void;
+    /**
+     * Fired at most once per region, before `transcribe` resolves.
+     * Implementations that always cover the full audio never fire it.
+     */
+    onUntranscribedRegion?: (region: UntranscribedRegion) => void;
     transcribe(audio: Blob, options?: TranscriberOptions): Promise<Document>;
 }
 //# sourceMappingURL=Transcriber.d.ts.map

package/dist/modules/transcription/Transcriber.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"Transcriber.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/Transcriber.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD,MAAM,WAAW,kBAAkB;IACjC,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED;;;;;;GAMG;AACH,MAAM,MAAM,wBAAwB,GAChC;IAAE,KAAK,EAAE,SAAS,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAA;CAAE,GACtC;IAAE,KAAK,EAAE,WAAW,CAAC;IAAC,QAAQ,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAE9C,MAAM,WAAW,WAAW;IAC1B,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,wBAAwB,KAAK,IAAI,CAAC;IACvD,UAAU,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,kBAAkB,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC;CAC1E"}
+{"version":3,"file":"Transcriber.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/Transcriber.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAC;AAExD,MAAM,WAAW,kBAAkB;IACjC,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED;;;;;;GAMG;AACH,MAAM,MAAM,wBAAwB,GAChC;IAAE,KAAK,EAAE,SAAS,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAA;CAAE,GACtC;IAAE,KAAK,EAAE,WAAW,CAAC;IAAC,QAAQ,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAE9C;;;GAGG;AACH,MAAM,WAAW,mBAAmB;IAClC,YAAY,EAAE,MAAM,CAAC;IACrB,UAAU,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,WAAW;IAC1B,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,wBAAwB,KAAK,IAAI,CAAC;IACvD;;;OAGG;IACH,qBAAqB,CAAC,EAAE,CAAC,MAAM,EAAE,mBAAmB,KAAK,IAAI,CAAC;IAC9D,UAAU,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,kBAAkB,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC;CAC1E"}

package/dist/modules/transcription/TransformersRuntime.d.ts

@@ -0,0 +1,50 @@
+import type * as Transformers from '@huggingface/transformers';
+import type { ModelFileCache } from '../transcription/ModelFileCache';
+/**
+ * The inference library, loaded on first use and configured once.
+ *
+ * It is reached through a dynamic import because a static one makes the
+ * whole ONNX runtime a dependency of merely *naming* a transcriber: the
+ * package's entry point re-exports this module, so a consumer that
+ * imports a document node pulls the runtime with it. In a browser that
+ * is a large bundle nobody asked for; in Node it is a native binding
+ * loaded — and on musl, failed — for code that never runs inference.
+ *
+ * The environment settings live here rather than at the call site
+ * because they have to be applied before the first pipeline is built and
+ * exactly once. Holding the module and the settings together is what
+ * makes "loaded" and "configured" the same event.
+ *
+ * A failed load is not cached: the import can fail on a dead network the
+ * same way the weights can, and a caller retrying deserves to reach the
+ * network again.
+ */
+export declare class TransformersRuntime {
+    private readonly modelFileCache?;
+    private modulePromise;
+    constructor(modelFileCache?: ModelFileCache | undefined);
+    /**
+     * Resolves with the library, configured. Concurrent callers share one
+     * load, and the settings are applied before any of them sees it.
+     */
+    load(): Promise<typeof Transformers>;
+    private configure;
+    /**
+     * Takes over where the downloaded model files are kept.
+     *
+     * Left to itself the library picks a store from the environment and
+     * discards a refused write with nothing but a console warning, so a
+     * caller that wants to know when the files did not survive the run
+     * has to own the store. Without one, that default stands.
+     */
+    private installModelFileCache;
+    /**
+     * Raises the WASM thread count above ORT's built-in ceiling when the
+     * runtime supports it. ORT locks the count to 1 without cross-origin
+     * isolation, and even with isolation its own default caps at 4 no matter
+     * the machine — Whisper's encoder scales past that. On a WebGPU run this
+     * has no effect; the WASM backend is not touched.
+     */
+    private configureWasmThreading;
+}
+//# sourceMappingURL=TransformersRuntime.d.ts.map

package/dist/modules/transcription/TransformersRuntime.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"TransformersRuntime.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/TransformersRuntime.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,KAAK,YAAY,MAAM,2BAA2B,CAAC;AAC/D,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,uCAAuC,CAAC;AAE5E;;;;;;;;;;;;;;;;;;GAkBG;AACH,qBAAa,mBAAmB;IAGlB,OAAO,CAAC,QAAQ,CAAC,cAAc,CAAC;IAF5C,OAAO,CAAC,aAAa,CAA6C;gBAErC,cAAc,CAAC,EAAE,cAAc,YAAA;IAE5D;;;OAGG;IACH,IAAI,IAAI,OAAO,CAAC,OAAO,YAAY,CAAC;IAcpC,OAAO,CAAC,SAAS;IAMjB;;;;;;;OAOG;IACH,OAAO,CAAC,qBAAqB;IAM7B;;;;;;OAMG;IACH,OAAO,CAAC,sBAAsB;CAM/B"}