@mux/ai 0.35.0 → 0.37.0

package/README.md

@@ -124,6 +124,9 @@ const result = await editCaptions("your-asset-id", "track-id", {
   replacements: [
     { find: "Mucks", replace: "Mux" },
   ],
+  speakerReplacements: [
+    { find: "speaker_0", replace: "Alice" },
+  ],
 });
 
 console.log(result.totalReplacementCount); // 8

package/dist/{index-fU3hmKZu.d.ts → index-BupL84pY.d.ts}

@@ -2,7 +2,7 @@ import { z } from 'zod';
 import { createAnthropic } from '@ai-sdk/anthropic';
 import { createGoogleGenerativeAI } from '@ai-sdk/google';
 import { createOpenAI } from '@ai-sdk/openai';
-import { f as ScopedMuxAIOptions, I as ImageSubmissionMode, i as TokenUsage, M as MuxAIOptions, b as ChunkingStrategy, k as VideoEmbeddingsResult, H as Hotspot, v as Shot, j as ToneType, g as StorageAdapter } from './shots-BMbViFbZ.js';
+import { f as ScopedMuxAIOptions, I as ImageSubmissionMode, i as TokenUsage, M as MuxAIOptions, b as ChunkingStrategy, k as VideoEmbeddingsResult, H as Hotspot, v as Shot, j as ToneType, g as StorageAdapter } from './shots-CUI09Ohe.js';
 
 /**
  * Language Code Conversion Utilities
@@ -740,6 +740,12 @@ interface CaptionReplacement {
     /** Match `find` with case-sensitivity. Defaults to `true`. */
     caseSensitive?: boolean;
 }
+interface SpeakerReplacement {
+    /** Current speaker label, without the surrounding square brackets. */
+    find: string;
+    /** New speaker label, without the surrounding square brackets. */
+    replace: string;
+}
 interface ReplacementRecord {
     cueStartTime: number;
     before: string;
@@ -755,6 +761,8 @@ interface EditCaptionsOptions<P extends SupportedProvider = SupportedProvider> e
     autoCensorProfanity?: AutoCensorProfanityOptions;
     /** Static find/replace pairs (no LLM needed). */
     replacements?: CaptionReplacement[];
+    /** Replacements applied only to bracketed speaker labels at the start of cues. */
+    speakerReplacements?: SpeakerReplacement[];
     /** Delete the original track after creating the edited one. Defaults to true. */
     deleteOriginalTrack?: boolean;
     /**
@@ -795,6 +803,9 @@ interface EditCaptionsResult {
     replacements?: {
         replacements: ReplacementRecord[];
     };
+    speakerReplacements?: {
+        replacements: ReplacementRecord[];
+    };
     uploadedTrackId?: string;
     presignedUrl?: string;
     usage?: TokenUsage;
@@ -880,6 +891,14 @@ declare function applyReplacements(rawVtt: string, replacements: CaptionReplacem
     editedVtt: string;
     replacements: ReplacementRecord[];
 };
+/**
+ * Replaces bracketed speaker labels only when they appear at the start of a cue.
+ * Spoken text that happens to contain the same value is left unchanged.
+ */
+declare function applySpeakerReplacements(rawVtt: string, speakerReplacements: SpeakerReplacement[]): {
+    editedVtt: string;
+    replacements: ReplacementRecord[];
+};
 declare function editCaptions<P extends SupportedProvider = SupportedProvider>(assetId: string, trackId: string, options: EditCaptionsOptions<P>): Promise<EditCaptionsResult>;
 
 /** Configuration accepted by `generateEmbeddings`. */
@@ -1545,6 +1564,7 @@ type index_Question = Question;
 type index_QuestionAnswer = QuestionAnswer;
 type index_QuestionAnswerType = QuestionAnswerType;
 type index_ReplacementRecord = ReplacementRecord;
+type index_SpeakerReplacement = SpeakerReplacement;
 type index_StaticRenditionCleanupOutcome = StaticRenditionCleanupOutcome;
 type index_SummarizationOptions = SummarizationOptions;
 type index_SummarizationPromptOverrides = SummarizationPromptOverrides;
@@ -1559,6 +1579,7 @@ type index_TranslationResult = TranslationResult;
 declare const index_aggregateTokenUsage: typeof aggregateTokenUsage;
 declare const index_applyOverrideLists: typeof applyOverrideLists;
 declare const index_applyReplacements: typeof applyReplacements;
+declare const index_applySpeakerReplacements: typeof applySpeakerReplacements;
 declare const index_askQuestions: typeof askQuestions;
 declare const index_buildReplacementRegex: typeof buildReplacementRegex;
 declare const index_burnedInCaptionsSchema: typeof burnedInCaptionsSchema;
@@ -1588,7 +1609,7 @@ declare const index_translationSchema: typeof translationSchema;
 declare const index_validateNeverTranslateTerms: typeof validateNeverTranslateTerms;
 declare const index_verifyNeverTranslateTerms: typeof verifyNeverTranslateTerms;
 declare namespace index {
-  export { index_ASK_QUESTIONS_DEFAULT_MAX_ANSWER_OPTION_LENGTH as ASK_QUESTIONS_DEFAULT_MAX_ANSWER_OPTION_LENGTH, index_ASK_QUESTIONS_DEFAULT_MAX_FREE_FORM_ANSWER_LENGTH as ASK_QUESTIONS_DEFAULT_MAX_FREE_FORM_ANSWER_LENGTH, index_ASK_QUESTIONS_MAX_QUESTION_LENGTH as ASK_QUESTIONS_MAX_QUESTION_LENGTH, type index_AskQuestionsOptions as AskQuestionsOptions, type index_AskQuestionsResult as AskQuestionsResult, type index_AskQuestionsType as AskQuestionsType, type index_AudioTranslationOptions as AudioTranslationOptions, type index_AudioTranslationResult as AudioTranslationResult, type index_AutoCensorProfanityOptions as AutoCensorProfanityOptions, type index_BurnedInCaptionsAnalysis as BurnedInCaptionsAnalysis, type index_BurnedInCaptionsOptions as BurnedInCaptionsOptions, type index_BurnedInCaptionsPromptOverrides as BurnedInCaptionsPromptOverrides, type index_BurnedInCaptionsPromptSections as BurnedInCaptionsPromptSections, type index_BurnedInCaptionsResult as BurnedInCaptionsResult, type index_CaptionReplacement as CaptionReplacement, type index_CensorMode as CensorMode, type index_Chapter as Chapter, type index_ChapterSystemPromptSections as ChapterSystemPromptSections, type index_ChaptersOptions as ChaptersOptions, type index_ChaptersPromptOverrides as ChaptersPromptOverrides, type index_ChaptersPromptSections as ChaptersPromptSections, type index_ChaptersResult as ChaptersResult, type index_ChaptersType as ChaptersType, index_DEFAULT_DESCRIPTION_LENGTH as DEFAULT_DESCRIPTION_LENGTH, index_DEFAULT_SUMMARY_KEYWORD_LIMIT as DEFAULT_SUMMARY_KEYWORD_LIMIT, index_DEFAULT_TITLE_LENGTH as DEFAULT_TITLE_LENGTH, type index_EditCaptionsOptions as EditCaptionsOptions, type index_EditCaptionsResult as EditCaptionsResult, type index_EmbeddingsOptions as EmbeddingsOptions, type index_EmbeddingsResult as EmbeddingsResult, type index_EngagementInsightsOptions as EngagementInsightsOptions, type index_EngagementInsightsResult as EngagementInsightsResult, index_GOOGLE_VISION_LIKELIHOOD_TO_SCORE as GOOGLE_VISION_LIKELIHOOD_TO_SCORE, index_HIVE_SEXUAL_CATEGORIES as HIVE_SEXUAL_CATEGORIES, index_HIVE_VIOLENCE_CATEGORIES as HIVE_VIOLENCE_CATEGORIES, type index_HiveModerationOutput as HiveModerationOutput, type index_HiveModerationSource as HiveModerationSource, type index_ModerationOptions as ModerationOptions, type index_ModerationProvider as ModerationProvider, type index_ModerationResult as ModerationResult, type index_MomentInsight as MomentInsight, type index_OverallInsight as OverallInsight, type index_ProfanityDetectionPayload as ProfanityDetectionPayload, type index_Question as Question, type index_QuestionAnswer as QuestionAnswer, type index_QuestionAnswerType as QuestionAnswerType, type index_ReplacementRecord as ReplacementRecord, type index_StaticRenditionCleanupOutcome as StaticRenditionCleanupOutcome, type index_SummarizationOptions as SummarizationOptions, type index_SummarizationPromptOverrides as SummarizationPromptOverrides, type index_SummarizationPromptSections as SummarizationPromptSections, type index_SummaryAndTagsResult as SummaryAndTagsResult, type index_SummaryType as SummaryType, type index_ThumbnailModerationScore as ThumbnailModerationScore, type index_TranslationChunkingOptions as TranslationChunkingOptions, type index_TranslationOptions as TranslationOptions, type index_TranslationPayload as TranslationPayload, type index_TranslationResult as TranslationResult, index_aggregateTokenUsage as aggregateTokenUsage, index_applyOverrideLists as applyOverrideLists, index_applyReplacements as applyReplacements, index_askQuestions as askQuestions, index_buildReplacementRegex as buildReplacementRegex, index_burnedInCaptionsSchema as burnedInCaptionsSchema, index_censorVttContent as censorVttContent, index_chapterSchema as chapterSchema, index_chaptersSchema as chaptersSchema, index_createReplacer as createReplacer, index_editCaptions as editCaptions, index_extractTranscriptSegmentsForHotspots as extractTranscriptSegmentsForHotspots, index_generateChapters as generateChapters, index_generateEmbeddings as generateEmbeddings, index_generateEngagementInsights as generateEngagementInsights, index_generateVideoEmbeddings as generateVideoEmbeddings, index_getModerationScores as getModerationScores, index_getSummaryAndTags as getSummaryAndTags, index_hasBurnedInCaptions as hasBurnedInCaptions, index_mapShotsToHotspots as mapShotsToHotspots, index_normalizeTranslatedVtt as normalizeTranslatedVtt, index_profanityDetectionSchema as profanityDetectionSchema, index_questionAnswerSchema as questionAnswerSchema, index_shouldSplitChunkTranslationError as shouldSplitChunkTranslationError, index_summarySchema as summarySchema, index_transformCueText as transformCueText, index_translateAudio as translateAudio, index_translateCaptions as translateCaptions, index_translationSchema as translationSchema, index_validateNeverTranslateTerms as validateNeverTranslateTerms, index_verifyNeverTranslateTerms as verifyNeverTranslateTerms };
+  export { index_ASK_QUESTIONS_DEFAULT_MAX_ANSWER_OPTION_LENGTH as ASK_QUESTIONS_DEFAULT_MAX_ANSWER_OPTION_LENGTH, index_ASK_QUESTIONS_DEFAULT_MAX_FREE_FORM_ANSWER_LENGTH as ASK_QUESTIONS_DEFAULT_MAX_FREE_FORM_ANSWER_LENGTH, index_ASK_QUESTIONS_MAX_QUESTION_LENGTH as ASK_QUESTIONS_MAX_QUESTION_LENGTH, type index_AskQuestionsOptions as AskQuestionsOptions, type index_AskQuestionsResult as AskQuestionsResult, type index_AskQuestionsType as AskQuestionsType, type index_AudioTranslationOptions as AudioTranslationOptions, type index_AudioTranslationResult as AudioTranslationResult, type index_AutoCensorProfanityOptions as AutoCensorProfanityOptions, type index_BurnedInCaptionsAnalysis as BurnedInCaptionsAnalysis, type index_BurnedInCaptionsOptions as BurnedInCaptionsOptions, type index_BurnedInCaptionsPromptOverrides as BurnedInCaptionsPromptOverrides, type index_BurnedInCaptionsPromptSections as BurnedInCaptionsPromptSections, type index_BurnedInCaptionsResult as BurnedInCaptionsResult, type index_CaptionReplacement as CaptionReplacement, type index_CensorMode as CensorMode, type index_Chapter as Chapter, type index_ChapterSystemPromptSections as ChapterSystemPromptSections, type index_ChaptersOptions as ChaptersOptions, type index_ChaptersPromptOverrides as ChaptersPromptOverrides, type index_ChaptersPromptSections as ChaptersPromptSections, type index_ChaptersResult as ChaptersResult, type index_ChaptersType as ChaptersType, index_DEFAULT_DESCRIPTION_LENGTH as DEFAULT_DESCRIPTION_LENGTH, index_DEFAULT_SUMMARY_KEYWORD_LIMIT as DEFAULT_SUMMARY_KEYWORD_LIMIT, index_DEFAULT_TITLE_LENGTH as DEFAULT_TITLE_LENGTH, type index_EditCaptionsOptions as EditCaptionsOptions, type index_EditCaptionsResult as EditCaptionsResult, type index_EmbeddingsOptions as EmbeddingsOptions, type index_EmbeddingsResult as EmbeddingsResult, type index_EngagementInsightsOptions as EngagementInsightsOptions, type index_EngagementInsightsResult as EngagementInsightsResult, index_GOOGLE_VISION_LIKELIHOOD_TO_SCORE as GOOGLE_VISION_LIKELIHOOD_TO_SCORE, index_HIVE_SEXUAL_CATEGORIES as HIVE_SEXUAL_CATEGORIES, index_HIVE_VIOLENCE_CATEGORIES as HIVE_VIOLENCE_CATEGORIES, type index_HiveModerationOutput as HiveModerationOutput, type index_HiveModerationSource as HiveModerationSource, type index_ModerationOptions as ModerationOptions, type index_ModerationProvider as ModerationProvider, type index_ModerationResult as ModerationResult, type index_MomentInsight as MomentInsight, type index_OverallInsight as OverallInsight, type index_ProfanityDetectionPayload as ProfanityDetectionPayload, type index_Question as Question, type index_QuestionAnswer as QuestionAnswer, type index_QuestionAnswerType as QuestionAnswerType, type index_ReplacementRecord as ReplacementRecord, type index_SpeakerReplacement as SpeakerReplacement, type index_StaticRenditionCleanupOutcome as StaticRenditionCleanupOutcome, type index_SummarizationOptions as SummarizationOptions, type index_SummarizationPromptOverrides as SummarizationPromptOverrides, type index_SummarizationPromptSections as SummarizationPromptSections, type index_SummaryAndTagsResult as SummaryAndTagsResult, type index_SummaryType as SummaryType, type index_ThumbnailModerationScore as ThumbnailModerationScore, type index_TranslationChunkingOptions as TranslationChunkingOptions, type index_TranslationOptions as TranslationOptions, type index_TranslationPayload as TranslationPayload, type index_TranslationResult as TranslationResult, index_aggregateTokenUsage as aggregateTokenUsage, index_applyOverrideLists as applyOverrideLists, index_applyReplacements as applyReplacements, index_applySpeakerReplacements as applySpeakerReplacements, index_askQuestions as askQuestions, index_buildReplacementRegex as buildReplacementRegex, index_burnedInCaptionsSchema as burnedInCaptionsSchema, index_censorVttContent as censorVttContent, index_chapterSchema as chapterSchema, index_chaptersSchema as chaptersSchema, index_createReplacer as createReplacer, index_editCaptions as editCaptions, index_extractTranscriptSegmentsForHotspots as extractTranscriptSegmentsForHotspots, index_generateChapters as generateChapters, index_generateEmbeddings as generateEmbeddings, index_generateEngagementInsights as generateEngagementInsights, index_generateVideoEmbeddings as generateVideoEmbeddings, index_getModerationScores as getModerationScores, index_getSummaryAndTags as getSummaryAndTags, index_hasBurnedInCaptions as hasBurnedInCaptions, index_mapShotsToHotspots as mapShotsToHotspots, index_normalizeTranslatedVtt as normalizeTranslatedVtt, index_profanityDetectionSchema as profanityDetectionSchema, index_questionAnswerSchema as questionAnswerSchema, index_shouldSplitChunkTranslationError as shouldSplitChunkTranslationError, index_summarySchema as summarySchema, index_transformCueText as transformCueText, index_translateAudio as translateAudio, index_translateCaptions as translateCaptions, index_translationSchema as translationSchema, index_validateNeverTranslateTerms as validateNeverTranslateTerms, index_verifyNeverTranslateTerms as verifyNeverTranslateTerms };
 }
 
-export { type MomentInsight as $, ASK_QUESTIONS_DEFAULT_MAX_ANSWER_OPTION_LENGTH as A, type BurnedInCaptionsAnalysis as B, type CaptionReplacement as C, type ChaptersPromptOverrides as D, type ChaptersPromptSections as E, type ChaptersResult as F, type ChaptersType as G, DEFAULT_DESCRIPTION_LENGTH as H, type ISO639_1 as I, DEFAULT_SUMMARY_KEYWORD_LIMIT as J, DEFAULT_TITLE_LENGTH as K, type LanguageCodePair as L, type EditCaptionsOptions as M, type EditCaptionsResult as N, type EmbeddingsOptions as O, type EmbeddingsResult as P, type EngagementInsightsOptions as Q, type EngagementInsightsResult as R, type SupportedISO639_1 as S, GOOGLE_VISION_LIKELIHOOD_TO_SCORE as T, HIVE_SEXUAL_CATEGORIES as U, HIVE_VIOLENCE_CATEGORIES as V, type HiveModerationOutput as W, type HiveModerationSource as X, type ModerationOptions as Y, type ModerationProvider as Z, type ModerationResult as _, type ISO639_3 as a, type OverallInsight as a0, type ProfanityDetectionPayload as a1, type Question as a2, type QuestionAnswer as a3, type QuestionAnswerType as a4, type ReplacementRecord as a5, type StaticRenditionCleanupOutcome as a6, type SummarizationOptions as a7, type SummarizationPromptOverrides as a8, type SummarizationPromptSections as a9, mapShotsToHotspots as aA, normalizeTranslatedVtt as aB, profanityDetectionSchema as aC, questionAnswerSchema as aD, shouldSplitChunkTranslationError as aE, summarySchema as aF, transformCueText as aG, translateAudio as aH, translateCaptions as aI, translationSchema as aJ, validateNeverTranslateTerms as aK, verifyNeverTranslateTerms as aL, type SummaryAndTagsResult as aa, type SummaryType as ab, type ThumbnailModerationScore as ac, type TranslationChunkingOptions as ad, type TranslationOptions as ae, type TranslationPayload as af, type TranslationResult as ag, aggregateTokenUsage as ah, applyOverrideLists as ai, applyReplacements as aj, askQuestions as ak, buildReplacementRegex as al, burnedInCaptionsSchema as am, censorVttContent as an, chapterSchema as ao, chaptersSchema as ap, createReplacer as aq, editCaptions as ar, extractTranscriptSegmentsForHotspots as as, generateChapters as at, generateEmbeddings as au, generateEngagementInsights as av, generateVideoEmbeddings as aw, getModerationScores as ax, getSummaryAndTags as ay, hasBurnedInCaptions as az, type SupportedISO639_3 as b, getLanguageName as c, isValidISO639_1 as d, isValidISO639_3 as e, toISO639_3 as f, getLanguageCodePair as g, index as h, isUndeterminedLanguageCode as i, ASK_QUESTIONS_DEFAULT_MAX_FREE_FORM_ANSWER_LENGTH as j, ASK_QUESTIONS_MAX_QUESTION_LENGTH as k, type AskQuestionsOptions as l, type AskQuestionsResult as m, type AskQuestionsType as n, type AudioTranslationOptions as o, type AudioTranslationResult as p, type AutoCensorProfanityOptions as q, type BurnedInCaptionsOptions as r, type BurnedInCaptionsPromptOverrides as s, toISO639_1 as t, type BurnedInCaptionsPromptSections as u, type BurnedInCaptionsResult as v, type CensorMode as w, type Chapter as x, type ChapterSystemPromptSections as y, type ChaptersOptions as z };
+export { type MomentInsight as $, ASK_QUESTIONS_DEFAULT_MAX_ANSWER_OPTION_LENGTH as A, type BurnedInCaptionsAnalysis as B, type CaptionReplacement as C, type ChaptersPromptOverrides as D, type ChaptersPromptSections as E, type ChaptersResult as F, type ChaptersType as G, DEFAULT_DESCRIPTION_LENGTH as H, type ISO639_1 as I, DEFAULT_SUMMARY_KEYWORD_LIMIT as J, DEFAULT_TITLE_LENGTH as K, type LanguageCodePair as L, type EditCaptionsOptions as M, type EditCaptionsResult as N, type EmbeddingsOptions as O, type EmbeddingsResult as P, type EngagementInsightsOptions as Q, type EngagementInsightsResult as R, type SupportedISO639_1 as S, GOOGLE_VISION_LIKELIHOOD_TO_SCORE as T, HIVE_SEXUAL_CATEGORIES as U, HIVE_VIOLENCE_CATEGORIES as V, type HiveModerationOutput as W, type HiveModerationSource as X, type ModerationOptions as Y, type ModerationProvider as Z, type ModerationResult as _, type ISO639_3 as a, type OverallInsight as a0, type ProfanityDetectionPayload as a1, type Question as a2, type QuestionAnswer as a3, type QuestionAnswerType as a4, type ReplacementRecord as a5, type SpeakerReplacement as a6, type StaticRenditionCleanupOutcome as a7, type SummarizationOptions as a8, type SummarizationPromptOverrides as a9, getSummaryAndTags as aA, hasBurnedInCaptions as aB, mapShotsToHotspots as aC, normalizeTranslatedVtt as aD, profanityDetectionSchema as aE, questionAnswerSchema as aF, shouldSplitChunkTranslationError as aG, summarySchema as aH, transformCueText as aI, translateAudio as aJ, translateCaptions as aK, translationSchema as aL, validateNeverTranslateTerms as aM, verifyNeverTranslateTerms as aN, type SummarizationPromptSections as aa, type SummaryAndTagsResult as ab, type SummaryType as ac, type ThumbnailModerationScore as ad, type TranslationChunkingOptions as ae, type TranslationOptions as af, type TranslationPayload as ag, type TranslationResult as ah, aggregateTokenUsage as ai, applyOverrideLists as aj, applyReplacements as ak, applySpeakerReplacements as al, askQuestions as am, buildReplacementRegex as an, burnedInCaptionsSchema as ao, censorVttContent as ap, chapterSchema as aq, chaptersSchema as ar, createReplacer as as, editCaptions as at, extractTranscriptSegmentsForHotspots as au, generateChapters as av, generateEmbeddings as aw, generateEngagementInsights as ax, generateVideoEmbeddings as ay, getModerationScores as az, type SupportedISO639_3 as b, getLanguageName as c, isValidISO639_1 as d, isValidISO639_3 as e, toISO639_3 as f, getLanguageCodePair as g, index as h, isUndeterminedLanguageCode as i, ASK_QUESTIONS_DEFAULT_MAX_FREE_FORM_ANSWER_LENGTH as j, ASK_QUESTIONS_MAX_QUESTION_LENGTH as k, type AskQuestionsOptions as l, type AskQuestionsResult as m, type AskQuestionsType as n, type AudioTranslationOptions as o, type AudioTranslationResult as p, type AutoCensorProfanityOptions as q, type BurnedInCaptionsOptions as r, type BurnedInCaptionsPromptOverrides as s, toISO639_1 as t, type BurnedInCaptionsPromptSections as u, type BurnedInCaptionsResult as v, type CensorMode as w, type Chapter as x, type ChapterSystemPromptSections as y, type ChaptersOptions as z };

package/dist/{index-SjpXUdXl.d.ts → index-C2Pk7bbF.d.ts}

@@ -1,4 +1,4 @@
-import { l as WorkflowCredentialsInput, n as WorkflowScope, A as AssetTextTrack, d as MuxAsset, T as TextChunk, b as ChunkingStrategy, q as CompletedShotsResult, r as ErroredShotsResult, H as Hotspot, s as HotspotOptions, t as HotspotResponse, u as PendingShotsResult, v as Shot, w as ShotRequestOptions, x as ShotsResult, y as WaitForShotsOptions, z as getHotspotsForAsset, B as getHotspotsForPlaybackId, D as getHotspotsForVideo, F as getShotsForAsset, G as requestShotsForAsset, J as waitForShotsForAsset } from './shots-BMbViFbZ.js';
+import { l as WorkflowCredentialsInput, n as WorkflowScope, A as AssetTextTrack, d as MuxAsset, T as TextChunk, b as ChunkingStrategy, q as CompletedShotsResult, r as ErroredShotsResult, H as Hotspot, s as HotspotOptions, t as HotspotResponse, u as PendingShotsResult, v as Shot, w as ShotRequestOptions, x as ShotsResult, y as WaitForShotsOptions, z as getHotspotsForAsset, B as getHotspotsForPlaybackId, D as getHotspotsForVideo, F as getShotsForAsset, G as requestShotsForAsset, J as waitForShotsForAsset } from './shots-CUI09Ohe.js';
 
 interface HeatmapOptions {
     /** Time window for results, e.g., '7:days' (default: '7:days') */

package/dist/index.d.ts

@@ -1,15 +1,15 @@
-export { I as ISO639_1, a as ISO639_3, L as LanguageCodePair, S as SupportedISO639_1, b as SupportedISO639_3, g as getLanguageCodePair, c as getLanguageName, i as isUndeterminedLanguageCode, d as isValidISO639_1, e as isValidISO639_3, t as toISO639_1, f as toISO639_3, h as workflows } from './index-fU3hmKZu.js';
-import { W as WorkflowCredentials, S as StoragePutObjectInput, a as StoragePresignGetObjectInput } from './shots-BMbViFbZ.js';
-export { A as AssetTextTrack, C as ChunkEmbedding, b as ChunkingStrategy, E as Encrypted, c as EncryptedPayload, I as ImageSubmissionMode, M as MuxAIOptions, d as MuxAsset, P as PlaybackAsset, e as PlaybackPolicy, f as ScopedMuxAIOptions, g as StorageAdapter, T as TextChunk, h as TokenChunkingConfig, i as TokenUsage, j as ToneType, U as UsageMetadata, V as VTTChunkingConfig, k as VideoEmbeddingsResult, l as WorkflowCredentialsInput, m as WorkflowMuxClient, n as WorkflowScope, o as decryptFromWorkflow, p as encryptForWorkflow } from './shots-BMbViFbZ.js';
+export { I as ISO639_1, a as ISO639_3, L as LanguageCodePair, S as SupportedISO639_1, b as SupportedISO639_3, g as getLanguageCodePair, c as getLanguageName, i as isUndeterminedLanguageCode, d as isValidISO639_1, e as isValidISO639_3, t as toISO639_1, f as toISO639_3, h as workflows } from './index-BupL84pY.js';
+import { W as WorkflowCredentials, S as StoragePutObjectInput, a as StoragePresignGetObjectInput } from './shots-CUI09Ohe.js';
+export { A as AssetTextTrack, C as ChunkEmbedding, b as ChunkingStrategy, E as Encrypted, c as EncryptedPayload, I as ImageSubmissionMode, M as MuxAIOptions, d as MuxAsset, P as PlaybackAsset, e as PlaybackPolicy, f as ScopedMuxAIOptions, g as StorageAdapter, T as TextChunk, h as TokenChunkingConfig, i as TokenUsage, j as ToneType, U as UsageMetadata, V as VTTChunkingConfig, k as VideoEmbeddingsResult, l as WorkflowCredentialsInput, m as WorkflowMuxClient, n as WorkflowScope, o as decryptFromWorkflow, p as encryptForWorkflow } from './shots-CUI09Ohe.js';
 import { WORKFLOW_SERIALIZE, WORKFLOW_DESERIALIZE } from '@workflow/serde';
-export { i as primitives } from './index-SjpXUdXl.js';
+export { i as primitives } from './index-C2Pk7bbF.js';
 import 'zod';
 import '@ai-sdk/anthropic';
 import '@ai-sdk/google';
 import '@ai-sdk/openai';
 import '@mux/mux-node';
 
-var version = "0.35.0";
+var version = "0.37.0";
 
 type MuxAiErrorType = "validation_error" | "processing_error" | "timeout_error" | "content_policy_error";
 /**

package/dist/index.js

@@ -5,7 +5,7 @@ var __export = (target, all) => {
 };
 
 // package.json
-var version = "0.35.0";
+var version = "0.37.0";
 
 // src/lib/language-codes.ts
 var ISO639_1_TO_3 = {
@@ -2534,9 +2534,12 @@ function toPlaybackAsset(asset) {
   const { id: playbackId, policy } = getPlaybackId(asset);
   return { asset, playbackId, policy };
 }
-async function getPlaybackIdForAsset(assetId, credentials) {
+async function getPlaybackIdForAsset(assetId, credentials, assetSnapshot) {
   "use step";
-  const asset = await getMuxAsset(assetId, credentials);
+  const asset = assetSnapshot ?? await getMuxAsset(assetId, credentials);
+  if (asset.id !== assetId) {
+    throw new MuxAiError("Asset snapshot does not match the requested asset ID.", { type: "validation_error" });
+  }
   return toPlaybackAsset(asset);
 }
 async function getMuxAsset(assetId, credentials) {
@@ -3019,6 +3022,7 @@ __export(workflows_exports, {
   aggregateTokenUsage: () => aggregateTokenUsage,
   applyOverrideLists: () => applyOverrideLists,
   applyReplacements: () => applyReplacements,
+  applySpeakerReplacements: () => applySpeakerReplacements,
   askQuestions: () => askQuestions,
   buildReplacementRegex: () => buildReplacementRegex,
   burnedInCaptionsSchema: () => burnedInCaptionsSchema,
@@ -3990,7 +3994,7 @@ async function askQuestionsInternal(assetId, questions, options, collectedUsage)
     model,
     provider
   });
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options?.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const isAudioOnly = isAudioOnlyAsset(assetData);
   const effectiveScope = hasWorkflowScopeBoundaries(scope) ? scope : void 0;
@@ -4344,7 +4348,7 @@ async function hasBurnedInCaptionsInternal(assetId, options, collectedUsage) {
     model,
     provider
   });
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const storyboardScope = resolveRenderableVideoScope(
     scope,
@@ -4609,7 +4613,7 @@ async function generateChaptersInternal(assetId, options, collectedUsage) {
     model,
     provider
   });
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const effectiveScope = hasWorkflowScopeBoundaries(scope) ? scope : void 0;
   const resolvedScope = effectiveScope ? resolveWorkflowScope(effectiveScope, assetDurationSeconds) : void 0;
@@ -4959,6 +4963,27 @@ function applyReplacements(rawVtt, replacements) {
   });
   return { editedVtt, replacements: records };
 }
+function applySpeakerReplacements(rawVtt, speakerReplacements) {
+  const replacementsByLabel = new Map(
+    speakerReplacements.filter((replacement) => replacement.find.length > 0).map((replacement) => [replacement.find, replacement.replace])
+  );
+  if (replacementsByLabel.size === 0) {
+    return { editedVtt: rawVtt, replacements: [] };
+  }
+  const records = [];
+  const editedVtt = transformCueTextBlocks(rawVtt, (cueText, cueStartTime) => {
+    return cueText.replace(/^(\s*)\[([^\]\r\n]+)\](?=\s|$)/, (marker, leadingWhitespace, label) => {
+      const replacement = replacementsByLabel.get(label);
+      if (replacement === void 0)
+        return marker;
+      const before = `[${label}]`;
+      const after = `[${replacement}]`;
+      records.push({ cueStartTime, before, after });
+      return `${leadingWhitespace}${after}`;
+    });
+  });
+  return { editedVtt, replacements: records };
+}
 async function identifyProfanityWithAI({
   plainText,
   provider,
@@ -5061,6 +5086,7 @@ async function editCaptionsInternal(assetId, trackId, options, collectedUsage) {
     model,
     autoCensorProfanity: autoCensorOption,
     replacements: replacementsOption,
+    speakerReplacements: speakerReplacementsOption,
     deleteOriginalTrack,
     uploadToS3: uploadToS3Option,
     uploadToMux: uploadToMuxOption,
@@ -5073,8 +5099,9 @@ async function editCaptionsInternal(assetId, trackId, options, collectedUsage) {
   } = options;
   const hasAutoCensor = !!autoCensorOption;
   const hasReplacements = !!replacementsOption && replacementsOption.length > 0;
-  if (!hasAutoCensor && !hasReplacements) {
-    throw new MuxAiError("At least one of autoCensorProfanity or replacements must be provided.", { type: "validation_error" });
+  const hasSpeakerReplacements = !!speakerReplacementsOption && speakerReplacementsOption.length > 0;
+  if (!hasAutoCensor && !hasReplacements && !hasSpeakerReplacements) {
+    throw new MuxAiError("At least one of autoCensorProfanity, replacements, or speakerReplacements must be provided.", { type: "validation_error" });
   }
   if (autoCensorOption && !provider) {
     throw new MuxAiError("provider is required when using autoCensorProfanity.", { type: "validation_error" });
@@ -5082,6 +5109,16 @@ async function editCaptionsInternal(assetId, trackId, options, collectedUsage) {
   const deleteOriginal = deleteOriginalTrack !== false;
   const uploadToMux = uploadToMuxOption !== false;
   const uploadToS3 = uploadToS3Option || uploadToMux;
+  for (const replacement of speakerReplacementsOption ?? []) {
+    const hasInvalidFindCharacter = replacement.find.includes("[") || replacement.find.includes("]") || /[\r\n]/.test(replacement.find);
+    const hasInvalidReplaceCharacter = replacement.replace.includes("[") || replacement.replace.includes("]") || /[\r\n]/.test(replacement.replace);
+    if (!replacement.find || !replacement.replace || hasInvalidFindCharacter || hasInvalidReplaceCharacter) {
+      throw new MuxAiError(
+        "Speaker replacement labels must be non-empty and must not contain square brackets or newlines.",
+        { type: "validation_error" }
+      );
+    }
+  }
   const s3Endpoint = providedS3Endpoint ?? env_default.S3_ENDPOINT;
   const s3Region = providedS3Region ?? env_default.S3_REGION ?? "auto";
   const s3Bucket = providedS3Bucket ?? env_default.S3_BUCKET;
@@ -5093,7 +5130,7 @@ async function editCaptionsInternal(assetId, trackId, options, collectedUsage) {
       { type: "validation_error" }
     );
   }
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const signingContext = await resolveMuxSigningContext(credentials);
   if (policy === "signed" && !signingContext) {
@@ -5163,6 +5200,16 @@ async function editCaptionsInternal(assetId, trackId, options, collectedUsage) {
     totalReplacementCount += censorReplacements.length;
     autoCensorResult = { replacements: censorReplacements };
   }
+  let speakerReplacementsResult;
+  if (speakerReplacementsOption && speakerReplacementsOption.length > 0) {
+    const { editedVtt: afterSpeakerReplacements, replacements: speakerReplacements } = applySpeakerReplacements(
+      editedVtt,
+      speakerReplacementsOption
+    );
+    editedVtt = afterSpeakerReplacements;
+    totalReplacementCount += speakerReplacements.length;
+    speakerReplacementsResult = { replacements: speakerReplacements };
+  }
   let replacementsResult;
   if (replacementsOption && replacementsOption.length > 0) {
     const { editedVtt: afterReplacements, replacements: staticReplacements } = applyReplacements(editedVtt, replacementsOption);
@@ -5225,6 +5272,7 @@ async function editCaptionsInternal(assetId, trackId, options, collectedUsage) {
     totalReplacementCount,
     autoCensorProfanity: autoCensorResult,
     replacements: replacementsResult,
+    speakerReplacements: speakerReplacementsResult,
     uploadedTrackId,
     presignedUrl,
     usage: usageWithMetadata,
@@ -5289,7 +5337,7 @@ async function generateEmbeddingsInternal(assetId, options, collectedUsage) {
     scope
   } = options;
   const embeddingModel = resolveEmbeddingModelConfig({ ...options, provider, model });
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const effectiveScope = hasWorkflowScopeBoundaries(scope) ? scope : void 0;
   if (effectiveScope) {
@@ -5763,7 +5811,7 @@ async function generateEngagementInsightsInternal(assetId, options, collectedUsa
     model,
     provider
   });
-  const { asset, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(asset);
   if (!assetDurationSeconds) {
     throw new MuxAiError(`Asset ${assetId} has no valid duration.`, { type: "validation_error" });
@@ -6480,7 +6528,7 @@ async function getModerationScores(assetId, options = {}) {
     scope
   } = options;
   const credentials = providedCredentials;
-  const { asset, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const videoTrackDurationSeconds = getVideoTrackDurationSecondsFromAsset(asset);
   const videoTrackFps = getVideoTrackMaxFrameRateFromAsset(asset);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(asset);
@@ -7094,7 +7142,7 @@ async function getSummaryAndTagsInternal(assetId, options, collectedUsage) {
     provider
   });
   const workflowCredentials = credentials;
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, workflowCredentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, workflowCredentials, options?.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const isAudioOnly = isAudioOnlyAsset(assetData);
   const effectiveScope = hasWorkflowScopeBoundaries(scope) ? scope : void 0;
@@ -7541,7 +7589,7 @@ async function translateAudio(assetId, toLanguageCode, options = {}) {
   if (uploadToS3 && (!s3Endpoint || !s3Bucket || !effectiveStorageAdapter && (!s3AccessKeyId || !s3SecretAccessKey))) {
     throw new MuxAiError("Storage configuration is required for uploading. Provide s3Endpoint and s3Bucket. If no storageAdapter is supplied, also provide s3AccessKeyId and s3SecretAccessKey in options or set S3_ENDPOINT, S3_BUCKET, S3_ACCESS_KEY_ID, and S3_SECRET_ACCESS_KEY environment variables.", { type: "validation_error" });
   }
-  const { asset: initialAsset, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: initialAsset, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(initialAsset);
   let currentAsset = initialAsset;
   let createdStaticRenditionId;
@@ -8411,7 +8459,7 @@ async function translateCaptionsInternal(assetId, trackId, toLanguageCode, optio
   if (uploadToS3 && (!s3Endpoint || !s3Bucket || !effectiveStorageAdapter && (!s3AccessKeyId || !s3SecretAccessKey))) {
     throw new MuxAiError("Storage configuration is required for uploading. Provide s3Endpoint and s3Bucket. If no storageAdapter is supplied, also provide s3AccessKeyId and s3SecretAccessKey in options or set S3_ENDPOINT, S3_BUCKET, S3_ACCESS_KEY_ID, and S3_SECRET_ACCESS_KEY environment variables.", { type: "validation_error" });
   }
-  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials);
+  const { asset: assetData, playbackId, policy } = await getPlaybackIdForAsset(assetId, credentials, options.assetSnapshot);
   const assetDurationSeconds = getAssetDurationSecondsFromAsset(assetData);
   const signingContext = await resolveMuxSigningContext(credentials);
   if (policy === "signed" && !signingContext) {