@storyteller-platform/align 0.1.12 → 0.1.14

package/dist/align/getSentenceRanges.cjs

@@ -26,8 +26,10 @@ __export(getSentenceRanges_exports, {
 });
 module.exports = __toCommonJS(getSentenceRanges_exports);
 var import_itertools = require("itertools");
+var import_runes2 = require("runes2");
 var import_ffmpeg = require("../common/ffmpeg.cjs");
 var import_errorAlign = require("../errorAlign/errorAlign.cjs");
+var import_utils = require("../errorAlign/utils.cjs");
 var import_slugify = require("./slugify.cjs");
 function findStartTimestamp(matchStartIndex, transcription) {
   const entry = transcription.timeline.find(
@@ -44,21 +46,96 @@ function findEndTimestamp(matchEndIndex, transcription) {
   const entry = transcription.timeline.findLast(
     (entry2) => (entry2.startOffsetUtf16 ?? 0) < matchEndIndex
   );
-  return (entry == null ? void 0 : entry.endTime) ?? null;
+  if (!entry) return null;
+  return {
+    start: entry.startTime,
+    end: entry.endTime,
+    audiofile: entry.audiofile
+  };
 }
 function getAlignmentsForSentence(sentence, alignments) {
   const result = [];
+  let score = Math.floor(sentence.length / 2);
   let sentenceIndex = 0;
   for (const alignment of alignments) {
     if (sentenceIndex === sentence.length) break;
     if (alignment.opType !== "INSERT") {
       sentenceIndex += alignment.ref.length + (sentenceIndex === 0 ? 0 : 1);
     }
+    if (alignment.opType === "DELETE" || alignment.opType === "INSERT" && sentenceIndex > 0) {
+      score -= (alignment.ref ?? alignment.hyp).length + 1;
+    }
     result.push(alignment);
   }
-  return result;
+  return {
+    alignments: result,
+    score: result.some((a) => a.opType === "MATCH") ? score : -1
+  };
 }
-async function getSentenceRanges(startSentence, endSentence, transcription, sentences, chapterOffset, chapterEndOffset, locale) {
+function errorAlignWithNarrowing(refSentences, hyp, narrowStart, narrowEnd) {
+  const firstAttempt = (0, import_errorAlign.errorAlign)(refSentences.join("-"), hyp);
+  let alignmentIndex = 0;
+  let firstGood = 0;
+  if (narrowStart) {
+    for (const sentence of refSentences) {
+      const { alignments: sentenceAlignments, score } = getAlignmentsForSentence(sentence, firstAttempt.slice(alignmentIndex));
+      alignmentIndex += sentenceAlignments.length;
+      if (sentence === "" || score <= 0) {
+        firstGood++;
+      } else {
+        break;
+      }
+    }
+  }
+  const reversedFirstAttempt = firstAttempt.toReversed().map((a) => {
+    if (!a.ref) return a;
+    return new import_utils.Alignment(
+      a.opType,
+      a.refSlice,
+      a.hypSlice,
+      (0, import_runes2.runes)(a.ref).toReversed().join(""),
+      a.hyp,
+      a.leftCompound,
+      a.rightCompound
+    );
+  });
+  let lastGood = 0;
+  alignmentIndex = 0;
+  if (narrowEnd) {
+    for (const sentence of (0, import_utils.reversed)(refSentences)) {
+      const reversedSentence = (0, import_runes2.runes)(sentence).toReversed().join("");
+      const { alignments: sentenceAlignments, score } = getAlignmentsForSentence(
+        reversedSentence,
+        reversedFirstAttempt.slice(alignmentIndex)
+      );
+      alignmentIndex += sentenceAlignments.length;
+      if (sentence === "" || score <= 0) {
+        lastGood++;
+      } else {
+        break;
+      }
+    }
+  }
+  lastGood = refSentences.length - lastGood;
+  if (firstGood <= 1 && lastGood >= refSentences.length - 2) {
+    return {
+      alignments: firstAttempt,
+      slice: [0, refSentences.length]
+    };
+  }
+  const slice = [
+    Math.max(firstGood - 1, 0),
+    Math.min(refSentences.length, lastGood + 1)
+  ];
+  const { alignments, slice: narrowed } = errorAlignWithNarrowing(
+    refSentences.slice(...slice),
+    hyp,
+    narrowStart,
+    narrowEnd
+  );
+  return { alignments, slice: [slice[0] + narrowed[0], slice[0] + narrowed[1]] };
+}
+async function getSentenceRanges(transcription, sentences, chapterOffset, chapterEndOffset, locale) {
   const sentenceRanges = [];
   const fullTranscript = transcription.transcript;
   const chapterTranscript = fullTranscript.slice(
@@ -66,58 +143,105 @@ async function getSentenceRanges(startSentence, endSentence, transcription, sent
     chapterEndOffset
   );
   const { result: slugifiedChapterTranscript, mapping: transcriptMapping } = await (0, import_slugify.slugify)(chapterTranscript, locale);
+  const slugifiedChapterSentences = [];
+  for (const s of sentences) {
+    const { result } = await (0, import_slugify.slugify)(s, locale);
+    slugifiedChapterSentences.push(result);
+  }
+  let firstFoundSentence = 0;
+  let lastFoundSentence = sentences.length - 1;
   let chapterTranscriptEndIndex = chapterOffset;
-  let chapterSentenceIndex = startSentence;
+  let chapterSentenceIndex = 0;
   let slugifiedChapterTranscriptWindowStartIndex = 0;
-  while (chapterSentenceIndex < endSentence) {
-    const slugifiedChapterSentenceWindowList = [];
+  while (chapterSentenceIndex < slugifiedChapterSentences.length) {
+    let slugifiedChapterSentenceWindowList = [];
     let sentenceWindowLength = 0;
     let i = chapterSentenceIndex;
-    while (sentenceWindowLength < 5e3 && i < sentences.length) {
-      const { result: sentence } = await (0, import_slugify.slugify)(sentences[i], locale);
+    while (sentenceWindowLength < 5e3 && i < slugifiedChapterSentences.length) {
+      const sentence = slugifiedChapterSentences[i];
       slugifiedChapterSentenceWindowList.push(sentence);
       sentenceWindowLength += sentence.length;
       i++;
     }
-    const slugifiedChapterSentenceWindow = slugifiedChapterSentenceWindowList.join("-");
+    const remainingSlugifiedSentences = slugifiedChapterSentences.slice(i);
+    const remainingSlugifiedSentenceLength = remainingSlugifiedSentences.reduce(
+      (acc, s) => acc + s.length,
+      0
+    );
+    if (remainingSlugifiedSentenceLength < 5e3) {
+      slugifiedChapterSentenceWindowList.push(...remainingSlugifiedSentences);
+      sentenceWindowLength += remainingSlugifiedSentenceLength;
+      i = slugifiedChapterSentences.length;
+    }
     const slugifiedChapterTranscriptWindow = slugifiedChapterTranscript.slice(
       slugifiedChapterTranscriptWindowStartIndex,
       slugifiedChapterTranscriptWindowStartIndex + sentenceWindowLength * 1.2
     );
-    const alignments = (0, import_errorAlign.errorAlign)(
-      slugifiedChapterSentenceWindow,
-      slugifiedChapterTranscriptWindow
-    );
+    let alignments;
+    let slice = [0, slugifiedChapterSentenceWindowList.length - 1];
+    if (chapterSentenceIndex === 0 || i === sentences.length) {
+      const result = errorAlignWithNarrowing(
+        slugifiedChapterSentenceWindowList,
+        slugifiedChapterTranscriptWindow,
+        chapterSentenceIndex === 0,
+        i === sentences.length
+      );
+      alignments = result.alignments;
+      slice = result.slice;
+      if (chapterSentenceIndex === 0) {
+        firstFoundSentence = chapterSentenceIndex + slice[0];
+      }
+      if (i === sentences.length) {
+        lastFoundSentence = chapterSentenceIndex + slice[0] + slice[1] - 1;
+      }
+      slugifiedChapterSentenceWindowList = slugifiedChapterSentenceWindowList.slice(...slice);
+    } else {
+      alignments = (0, import_errorAlign.errorAlign)(
+        slugifiedChapterSentenceWindowList.join("-"),
+        slugifiedChapterTranscriptWindow
+      );
+    }
     let alignmentIndex = 0;
     let currentTranscriptWindowIndex = 0;
-    for (const [i2, slugifiedSentence] of (0, import_itertools.enumerate)(
+    for (const [j, slugifiedSentence] of (0, import_itertools.enumerate)(
       slugifiedChapterSentenceWindowList
     )) {
       if (!slugifiedSentence) continue;
-      const sentenceAlignments = getAlignmentsForSentence(
+      const { alignments: sentenceAlignments, score } = getAlignmentsForSentence(
         slugifiedSentence,
         alignments.slice(alignmentIndex)
       );
       const sentenceLengthInSlugifiedTranscript = sentenceAlignments.filter((a) => a.opType !== "DELETE").map((a) => a.hyp).join("-").length;
-      const start = findStartTimestamp(
-        chapterOffset + transcriptMapping.invert().map(
-          slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex,
-          1
-        ),
-        transcription
-      );
-      chapterTranscriptEndIndex = chapterOffset + transcriptMapping.invert().map(
-        slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex + sentenceLengthInSlugifiedTranscript,
-        -1
-      );
-      const end = findEndTimestamp(chapterTranscriptEndIndex, transcription);
-      if (start && end !== null) {
-        sentenceRanges.push({
-          id: i2 + chapterSentenceIndex,
-          start: start.start,
-          audiofile: start.audiofile,
-          end
-        });
+      if (score > 0) {
+        const start = findStartTimestamp(
+          chapterOffset + transcriptMapping.invert().map(
+            slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex,
+            1
+          ),
+          transcription
+        );
+        chapterTranscriptEndIndex = chapterOffset + transcriptMapping.invert().map(
+          slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex + sentenceLengthInSlugifiedTranscript,
+          -1
+        );
+        const end = findEndTimestamp(chapterTranscriptEndIndex, transcription);
+        if (start && end) {
+          if (start.audiofile !== end.audiofile) {
+            sentenceRanges.push({
+              id: j + chapterSentenceIndex + slice[0],
+              start: 0,
+              audiofile: end.audiofile,
+              end: end.end
+            });
+          } else {
+            sentenceRanges.push({
+              id: j + chapterSentenceIndex + slice[0],
+              start: start.start,
+              audiofile: start.audiofile,
+              end: end.end
+            });
+          }
+        }
       }
       alignmentIndex += sentenceAlignments.length;
       currentTranscriptWindowIndex += sentenceLengthInSlugifiedTranscript;
@@ -125,13 +249,18 @@ async function getSentenceRanges(startSentence, endSentence, transcription, sent
         currentTranscriptWindowIndex++;
       }
     }
-    chapterSentenceIndex += slugifiedChapterSentenceWindowList.length;
+    chapterSentenceIndex = i;
     slugifiedChapterTranscriptWindowStartIndex += currentTranscriptWindowIndex;
     if (slugifiedChapterTranscript[slugifiedChapterTranscriptWindowStartIndex] === "-") {
       slugifiedChapterTranscriptWindowStartIndex++;
     }
   }
-  return { sentenceRanges, transcriptionOffset: chapterTranscriptEndIndex };
+  return {
+    sentenceRanges,
+    transcriptionOffset: chapterTranscriptEndIndex,
+    firstFoundSentence,
+    lastFoundSentence
+  };
 }
 async function getLargestGap(trailing, leading) {
   const leadingGap = leading.start;

package/dist/align/getSentenceRanges.d.cts

@@ -13,10 +13,16 @@ type SentenceRange = {
     end: number;
     audiofile: string;
 };
-declare function findEndTimestamp(matchEndIndex: number, transcription: StorytellerTranscription): number | null;
-declare function getSentenceRanges(startSentence: number, endSentence: number, transcription: StorytellerTranscription, sentences: string[], chapterOffset: number, chapterEndOffset: number, locale: Intl.Locale): Promise<{
+declare function findEndTimestamp(matchEndIndex: number, transcription: StorytellerTranscription): {
+    start: number;
+    end: number;
+    audiofile: string;
+} | null;
+declare function getSentenceRanges(transcription: StorytellerTranscription, sentences: string[], chapterOffset: number, chapterEndOffset: number, locale: Intl.Locale): Promise<{
     sentenceRanges: SentenceRange[];
     transcriptionOffset: number;
+    firstFoundSentence: number;
+    lastFoundSentence: number;
 }>;
 declare function interpolateSentenceRanges(sentenceRanges: SentenceRange[], lastSentenceRange: SentenceRange | null): Promise<SentenceRange[]>;
 /**

package/dist/align/getSentenceRanges.d.ts

@@ -13,10 +13,16 @@ type SentenceRange = {
     end: number;
     audiofile: string;
 };
-declare function findEndTimestamp(matchEndIndex: number, transcription: StorytellerTranscription): number | null;
-declare function getSentenceRanges(startSentence: number, endSentence: number, transcription: StorytellerTranscription, sentences: string[], chapterOffset: number, chapterEndOffset: number, locale: Intl.Locale): Promise<{
+declare function findEndTimestamp(matchEndIndex: number, transcription: StorytellerTranscription): {
+    start: number;
+    end: number;
+    audiofile: string;
+} | null;
+declare function getSentenceRanges(transcription: StorytellerTranscription, sentences: string[], chapterOffset: number, chapterEndOffset: number, locale: Intl.Locale): Promise<{
     sentenceRanges: SentenceRange[];
     transcriptionOffset: number;
+    firstFoundSentence: number;
+    lastFoundSentence: number;
 }>;
 declare function interpolateSentenceRanges(sentenceRanges: SentenceRange[], lastSentenceRange: SentenceRange | null): Promise<SentenceRange[]>;
 /**

package/dist/align/getSentenceRanges.js

@@ -1,7 +1,9 @@
 import "../chunk-BIEQXUOY.js";
 import { enumerate } from "itertools";
+import { runes } from "runes2";
 import { getTrackDuration } from "../common/ffmpeg.js";
 import { errorAlign } from "../errorAlign/errorAlign.js";
+import { Alignment, reversed } from "../errorAlign/utils.js";
 import { slugify } from "./slugify.js";
 function findStartTimestamp(matchStartIndex, transcription) {
   const entry = transcription.timeline.find(
@@ -18,21 +20,96 @@ function findEndTimestamp(matchEndIndex, transcription) {
   const entry = transcription.timeline.findLast(
     (entry2) => (entry2.startOffsetUtf16 ?? 0) < matchEndIndex
   );
-  return (entry == null ? void 0 : entry.endTime) ?? null;
+  if (!entry) return null;
+  return {
+    start: entry.startTime,
+    end: entry.endTime,
+    audiofile: entry.audiofile
+  };
 }
 function getAlignmentsForSentence(sentence, alignments) {
   const result = [];
+  let score = Math.floor(sentence.length / 2);
   let sentenceIndex = 0;
   for (const alignment of alignments) {
     if (sentenceIndex === sentence.length) break;
     if (alignment.opType !== "INSERT") {
       sentenceIndex += alignment.ref.length + (sentenceIndex === 0 ? 0 : 1);
     }
+    if (alignment.opType === "DELETE" || alignment.opType === "INSERT" && sentenceIndex > 0) {
+      score -= (alignment.ref ?? alignment.hyp).length + 1;
+    }
     result.push(alignment);
   }
-  return result;
+  return {
+    alignments: result,
+    score: result.some((a) => a.opType === "MATCH") ? score : -1
+  };
 }
-async function getSentenceRanges(startSentence, endSentence, transcription, sentences, chapterOffset, chapterEndOffset, locale) {
+function errorAlignWithNarrowing(refSentences, hyp, narrowStart, narrowEnd) {
+  const firstAttempt = errorAlign(refSentences.join("-"), hyp);
+  let alignmentIndex = 0;
+  let firstGood = 0;
+  if (narrowStart) {
+    for (const sentence of refSentences) {
+      const { alignments: sentenceAlignments, score } = getAlignmentsForSentence(sentence, firstAttempt.slice(alignmentIndex));
+      alignmentIndex += sentenceAlignments.length;
+      if (sentence === "" || score <= 0) {
+        firstGood++;
+      } else {
+        break;
+      }
+    }
+  }
+  const reversedFirstAttempt = firstAttempt.toReversed().map((a) => {
+    if (!a.ref) return a;
+    return new Alignment(
+      a.opType,
+      a.refSlice,
+      a.hypSlice,
+      runes(a.ref).toReversed().join(""),
+      a.hyp,
+      a.leftCompound,
+      a.rightCompound
+    );
+  });
+  let lastGood = 0;
+  alignmentIndex = 0;
+  if (narrowEnd) {
+    for (const sentence of reversed(refSentences)) {
+      const reversedSentence = runes(sentence).toReversed().join("");
+      const { alignments: sentenceAlignments, score } = getAlignmentsForSentence(
+        reversedSentence,
+        reversedFirstAttempt.slice(alignmentIndex)
+      );
+      alignmentIndex += sentenceAlignments.length;
+      if (sentence === "" || score <= 0) {
+        lastGood++;
+      } else {
+        break;
+      }
+    }
+  }
+  lastGood = refSentences.length - lastGood;
+  if (firstGood <= 1 && lastGood >= refSentences.length - 2) {
+    return {
+      alignments: firstAttempt,
+      slice: [0, refSentences.length]
+    };
+  }
+  const slice = [
+    Math.max(firstGood - 1, 0),
+    Math.min(refSentences.length, lastGood + 1)
+  ];
+  const { alignments, slice: narrowed } = errorAlignWithNarrowing(
+    refSentences.slice(...slice),
+    hyp,
+    narrowStart,
+    narrowEnd
+  );
+  return { alignments, slice: [slice[0] + narrowed[0], slice[0] + narrowed[1]] };
+}
+async function getSentenceRanges(transcription, sentences, chapterOffset, chapterEndOffset, locale) {
   const sentenceRanges = [];
   const fullTranscript = transcription.transcript;
   const chapterTranscript = fullTranscript.slice(
@@ -40,58 +117,105 @@ async function getSentenceRanges(startSentence, endSentence, transcription, sent
     chapterEndOffset
   );
   const { result: slugifiedChapterTranscript, mapping: transcriptMapping } = await slugify(chapterTranscript, locale);
+  const slugifiedChapterSentences = [];
+  for (const s of sentences) {
+    const { result } = await slugify(s, locale);
+    slugifiedChapterSentences.push(result);
+  }
+  let firstFoundSentence = 0;
+  let lastFoundSentence = sentences.length - 1;
   let chapterTranscriptEndIndex = chapterOffset;
-  let chapterSentenceIndex = startSentence;
+  let chapterSentenceIndex = 0;
   let slugifiedChapterTranscriptWindowStartIndex = 0;
-  while (chapterSentenceIndex < endSentence) {
-    const slugifiedChapterSentenceWindowList = [];
+  while (chapterSentenceIndex < slugifiedChapterSentences.length) {
+    let slugifiedChapterSentenceWindowList = [];
     let sentenceWindowLength = 0;
     let i = chapterSentenceIndex;
-    while (sentenceWindowLength < 5e3 && i < sentences.length) {
-      const { result: sentence } = await slugify(sentences[i], locale);
+    while (sentenceWindowLength < 5e3 && i < slugifiedChapterSentences.length) {
+      const sentence = slugifiedChapterSentences[i];
       slugifiedChapterSentenceWindowList.push(sentence);
       sentenceWindowLength += sentence.length;
       i++;
     }
-    const slugifiedChapterSentenceWindow = slugifiedChapterSentenceWindowList.join("-");
+    const remainingSlugifiedSentences = slugifiedChapterSentences.slice(i);
+    const remainingSlugifiedSentenceLength = remainingSlugifiedSentences.reduce(
+      (acc, s) => acc + s.length,
+      0
+    );
+    if (remainingSlugifiedSentenceLength < 5e3) {
+      slugifiedChapterSentenceWindowList.push(...remainingSlugifiedSentences);
+      sentenceWindowLength += remainingSlugifiedSentenceLength;
+      i = slugifiedChapterSentences.length;
+    }
     const slugifiedChapterTranscriptWindow = slugifiedChapterTranscript.slice(
       slugifiedChapterTranscriptWindowStartIndex,
       slugifiedChapterTranscriptWindowStartIndex + sentenceWindowLength * 1.2
     );
-    const alignments = errorAlign(
-      slugifiedChapterSentenceWindow,
-      slugifiedChapterTranscriptWindow
-    );
+    let alignments;
+    let slice = [0, slugifiedChapterSentenceWindowList.length - 1];
+    if (chapterSentenceIndex === 0 || i === sentences.length) {
+      const result = errorAlignWithNarrowing(
+        slugifiedChapterSentenceWindowList,
+        slugifiedChapterTranscriptWindow,
+        chapterSentenceIndex === 0,
+        i === sentences.length
+      );
+      alignments = result.alignments;
+      slice = result.slice;
+      if (chapterSentenceIndex === 0) {
+        firstFoundSentence = chapterSentenceIndex + slice[0];
+      }
+      if (i === sentences.length) {
+        lastFoundSentence = chapterSentenceIndex + slice[0] + slice[1] - 1;
+      }
+      slugifiedChapterSentenceWindowList = slugifiedChapterSentenceWindowList.slice(...slice);
+    } else {
+      alignments = errorAlign(
+        slugifiedChapterSentenceWindowList.join("-"),
+        slugifiedChapterTranscriptWindow
+      );
+    }
     let alignmentIndex = 0;
     let currentTranscriptWindowIndex = 0;
-    for (const [i2, slugifiedSentence] of enumerate(
+    for (const [j, slugifiedSentence] of enumerate(
       slugifiedChapterSentenceWindowList
     )) {
       if (!slugifiedSentence) continue;
-      const sentenceAlignments = getAlignmentsForSentence(
+      const { alignments: sentenceAlignments, score } = getAlignmentsForSentence(
         slugifiedSentence,
         alignments.slice(alignmentIndex)
       );
       const sentenceLengthInSlugifiedTranscript = sentenceAlignments.filter((a) => a.opType !== "DELETE").map((a) => a.hyp).join("-").length;
-      const start = findStartTimestamp(
-        chapterOffset + transcriptMapping.invert().map(
-          slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex,
-          1
-        ),
-        transcription
-      );
-      chapterTranscriptEndIndex = chapterOffset + transcriptMapping.invert().map(
-        slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex + sentenceLengthInSlugifiedTranscript,
-        -1
-      );
-      const end = findEndTimestamp(chapterTranscriptEndIndex, transcription);
-      if (start && end !== null) {
-        sentenceRanges.push({
-          id: i2 + chapterSentenceIndex,
-          start: start.start,
-          audiofile: start.audiofile,
-          end
-        });
+      if (score > 0) {
+        const start = findStartTimestamp(
+          chapterOffset + transcriptMapping.invert().map(
+            slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex,
+            1
+          ),
+          transcription
+        );
+        chapterTranscriptEndIndex = chapterOffset + transcriptMapping.invert().map(
+          slugifiedChapterTranscriptWindowStartIndex + currentTranscriptWindowIndex + sentenceLengthInSlugifiedTranscript,
+          -1
+        );
+        const end = findEndTimestamp(chapterTranscriptEndIndex, transcription);
+        if (start && end) {
+          if (start.audiofile !== end.audiofile) {
+            sentenceRanges.push({
+              id: j + chapterSentenceIndex + slice[0],
+              start: 0,
+              audiofile: end.audiofile,
+              end: end.end
+            });
+          } else {
+            sentenceRanges.push({
+              id: j + chapterSentenceIndex + slice[0],
+              start: start.start,
+              audiofile: start.audiofile,
+              end: end.end
+            });
+          }
+        }
       }
       alignmentIndex += sentenceAlignments.length;
       currentTranscriptWindowIndex += sentenceLengthInSlugifiedTranscript;
@@ -99,13 +223,18 @@ async function getSentenceRanges(startSentence, endSentence, transcription, sent
         currentTranscriptWindowIndex++;
       }
     }
-    chapterSentenceIndex += slugifiedChapterSentenceWindowList.length;
+    chapterSentenceIndex = i;
     slugifiedChapterTranscriptWindowStartIndex += currentTranscriptWindowIndex;
     if (slugifiedChapterTranscript[slugifiedChapterTranscriptWindowStartIndex] === "-") {
       slugifiedChapterTranscriptWindowStartIndex++;
     }
   }
-  return { sentenceRanges, transcriptionOffset: chapterTranscriptEndIndex };
+  return {
+    sentenceRanges,
+    transcriptionOffset: chapterTranscriptEndIndex,
+    firstFoundSentence,
+    lastFoundSentence
+  };
 }
 async function getLargestGap(trailing, leading) {
   const leadingGap = leading.start;