@dxos/pipeline-transcription 0.0.0

package/dist/types/src/stages/correction.test.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"correction.test.d.ts","sourceRoot":"","sources":["../../../../src/stages/correction.test.ts"],"names":[],"mappings":""}

package/dist/types/src/stages/diarization.d.ts

@@ -0,0 +1,11 @@
+import { type ContentBlock } from '@dxos/types';
+import { type Stage } from '../Stage';
+export type DiarizationInput = {
+    window: ContentBlock.Transcript[];
+};
+/**
+ * Stage ◆ (interface only / deferred): label the speaker of the newest block.
+ * Conforms to the `Stage` contract and is inert until a diarization backend is wired.
+ */
+export declare const makeDiarizationStage: () => Stage<DiarizationInput>;
+//# sourceMappingURL=diarization.d.ts.map

package/dist/types/src/stages/diarization.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"diarization.d.ts","sourceRoot":"","sources":["../../../../src/stages/diarization.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,KAAK,YAAY,EAAE,MAAM,aAAa,CAAC;AAEhD,OAAO,EAAE,KAAK,KAAK,EAAc,MAAM,UAAU,CAAC;AAElD,MAAM,MAAM,gBAAgB,GAAG;IAAE,MAAM,EAAE,YAAY,CAAC,UAAU,EAAE,CAAA;CAAE,CAAC;AAErE;;;GAGG;AACH,eAAO,MAAM,oBAAoB,QAAO,KAAK,CAAC,gBAAgB,CAO5D,CAAC"}

package/dist/types/src/stages/extraction.d.ts

@@ -0,0 +1,40 @@
+import { type Ref } from '@dxos/echo';
+import { type ContentBlock } from '@dxos/types';
+import { type EntityLookup } from '../lookup';
+import { type Stage } from '../Stage';
+export type ExtractionInput = {
+    window: ContentBlock.Transcript[];
+};
+/**
+ * Deterministic proper-noun heuristic: capitalized word runs, minus sentence-initial stop-words.
+ * The LLM seam (see `run`) replaces this with model-driven NER.
+ */
+export declare const extractProperNouns: (text: string) => string[];
+/** Marker that a block's text already carries inline echo links, so extraction can skip it. */
+export declare const isLinked: (text: string) => boolean;
+/** Result of resolving a block's proper nouns against the {@link EntityLookup}. */
+export type ExtractedEntities = {
+    references: Ref.Ref<any>[];
+    quotes: {
+        quote: string;
+        id: string;
+    }[];
+    candidates: ContentBlock.Candidate[];
+};
+/** Resolve a block's proper nouns: matched → references/quotes, unmatched → candidates. */
+export declare const resolveEntities: (text: string, lookup?: EntityLookup) => Promise<ExtractedEntities>;
+/**
+ * Rewrite text with inline `[noun](echo:/<id>)` links for entities matched by `lookup`, so markdown
+ * consumers decorate them as dx-anchors. Reused by the live recording driver's post-process seam.
+ */
+export declare const linkEntities: (text: string, lookup: EntityLookup) => Promise<string>;
+/**
+ * Stage ②: surface proper nouns across the window, link those that match objects (via the injected
+ * {@link EntityLookup}), and report the rest as candidates the user can act on.
+ *
+ * Processes every not-yet-linked block in the window (not just the newest). Under `latest-wins`
+ * with a fast stream, earlier per-block invocations are interrupted, so the surviving invocation
+ * must cover the whole window — otherwise blocks behind the head never get extracted.
+ */
+export declare const makeExtractionStage: () => Stage<ExtractionInput>;
+//# sourceMappingURL=extraction.d.ts.map

package/dist/types/src/stages/extraction.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extraction.d.ts","sourceRoot":"","sources":["../../../../src/stages/extraction.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,KAAK,GAAG,EAAE,MAAM,YAAY,CAAC;AACtC,OAAO,EAAE,KAAK,YAAY,EAAE,MAAM,aAAa,CAAC;AAEhD,OAAO,EAAE,KAAK,YAAY,EAAE,MAAM,WAAW,CAAC;AAE9C,OAAO,EAAoB,KAAK,KAAK,EAAc,MAAM,UAAU,CAAC;AAEpE,MAAM,MAAM,eAAe,GAAG;IAAE,MAAM,EAAE,YAAY,CAAC,UAAU,EAAE,CAAA;CAAE,CAAC;AA4BpE;;;GAGG;AACH,eAAO,MAAM,kBAAkB,SAAU,MAAM,KAAG,MAAM,EAGvD,CAAC;AAEF,+FAA+F;AAC/F,eAAO,MAAM,QAAQ,SAAU,MAAM,KAAG,OAAmC,CAAC;AAE5E,mFAAmF;AACnF,MAAM,MAAM,iBAAiB,GAAG;IAC9B,UAAU,EAAE,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;IAC3B,MAAM,EAAE;QAAE,KAAK,EAAE,MAAM,CAAC;QAAC,EAAE,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;IACxC,UAAU,EAAE,YAAY,CAAC,SAAS,EAAE,CAAC;CACtC,CAAC;AAEF,2FAA2F;AAC3F,eAAO,MAAM,eAAe,SAAgB,MAAM,WAAW,YAAY,KAAG,OAAO,CAAC,iBAAiB,CA+BpG,CAAC;AAEF;;;GAGG;AACH,eAAO,MAAM,YAAY,SAAgB,MAAM,UAAU,YAAY,KAAG,OAAO,CAAC,MAAM,CAMrF,CAAC;AAEF;;;;;;;GAOG;AACH,eAAO,MAAM,mBAAmB,QAAO,KAAK,CAAC,eAAe,CAmC1D,CAAC"}

package/dist/types/src/stages/extraction.test.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=extraction.test.d.ts.map

package/dist/types/src/stages/extraction.test.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extraction.test.d.ts","sourceRoot":"","sources":["../../../../src/stages/extraction.test.ts"],"names":[],"mappings":""}

package/dist/types/src/stages/index.d.ts

@@ -0,0 +1,18 @@
+import { type Stage } from '../Stage';
+export * from './correction';
+export * from './correction-llm';
+export * from './extraction';
+export * from './summarization';
+export * from './translation';
+export * from './diarization';
+/**
+ * The default registry of stage factories keyed by stage id. Consumers build a stage list for a
+ * `PipelineConfig` by selecting enabled ids in config order.
+ */
+export declare const STAGE_FACTORIES: Readonly<Record<string, () => Stage<any, any>>>;
+/** Build the ordered, enabled stage list for a set of stage configs. */
+export declare const buildStages: (configs: readonly {
+    id: string;
+    enabled: boolean;
+}[]) => Stage<any, any>[];
+//# sourceMappingURL=index.d.ts.map

package/dist/types/src/stages/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/stages/index.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,KAAK,KAAK,EAAE,MAAM,UAAU,CAAC;AAOtC,cAAc,cAAc,CAAC;AAC7B,cAAc,kBAAkB,CAAC;AACjC,cAAc,cAAc,CAAC;AAC7B,cAAc,iBAAiB,CAAC;AAChC,cAAc,eAAe,CAAC;AAC9B,cAAc,eAAe,CAAC;AAE9B;;;GAGG;AACH,eAAO,MAAM,eAAe,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAM1E,CAAC;AAEH,wEAAwE;AACxE,eAAO,MAAM,WAAW,YAAa,SAAS;IAAE,EAAE,EAAE,MAAM,CAAC;IAAC,OAAO,EAAE,OAAO,CAAA;CAAE,EAAE,KAAG,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,EAUjG,CAAC"}

package/dist/types/src/stages/summarization.d.ts

@@ -0,0 +1,12 @@
+import { type ContentBlock } from '@dxos/types';
+import { type Stage } from '../Stage';
+export type SummarizationInput = {
+    window: ContentBlock.Transcript[];
+};
+/**
+ * Stage ④: maintain a cumulative summary. Fires on the silence edge, skips if a summary is already
+ * in flight. The deterministic implementation concatenates corrected utterances; the LLM seam
+ * (see `run`) replaces it with a cumulative model summary plus referent resolution.
+ */
+export declare const makeSummarizationStage: () => Stage<SummarizationInput>;
+//# sourceMappingURL=summarization.d.ts.map

package/dist/types/src/stages/summarization.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"summarization.d.ts","sourceRoot":"","sources":["../../../../src/stages/summarization.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,KAAK,YAAY,EAAE,MAAM,aAAa,CAAC;AAGhD,OAAO,EAAE,KAAK,KAAK,EAAc,MAAM,UAAU,CAAC;AAElD,MAAM,MAAM,kBAAkB,GAAG;IAAE,MAAM,EAAE,YAAY,CAAC,UAAU,EAAE,CAAA;CAAE,CAAC;AAEvE;;;;GAIG;AACH,eAAO,MAAM,sBAAsB,QAAO,KAAK,CAAC,kBAAkB,CAkBhE,CAAC"}

package/dist/types/src/stages/translation.d.ts

@@ -0,0 +1,11 @@
+import { type ContentBlock } from '@dxos/types';
+import { type Stage } from '../Stage';
+export type TranslationInput = {
+    window: ContentBlock.Transcript[];
+};
+/**
+ * Stage ③ (interface only / deferred): translate the newest block into `targetLanguage`.
+ * Conforms to the `Stage` contract and is inert until a translation backend is wired.
+ */
+export declare const makeTranslationStage: (_targetLanguage?: string) => Stage<TranslationInput>;
+//# sourceMappingURL=translation.d.ts.map

package/dist/types/src/stages/translation.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"translation.d.ts","sourceRoot":"","sources":["../../../../src/stages/translation.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,KAAK,YAAY,EAAE,MAAM,aAAa,CAAC;AAGhD,OAAO,EAAE,KAAK,KAAK,EAAc,MAAM,UAAU,CAAC;AAElD,MAAM,MAAM,gBAAgB,GAAG;IAAE,MAAM,EAAE,YAAY,CAAC,UAAU,EAAE,CAAA;CAAE,CAAC;AAErE;;;GAGG;AACH,eAAO,MAAM,oBAAoB,gCAA6B,KAAK,CAAC,gBAAgB,CAQlF,CAAC"}

package/dist/types/src/testing/conversation-parser.d.ts

@@ -0,0 +1,26 @@
+export type ConversationEntity = {
+    /** Type name as written in the fixture, e.g. `Person`, `Organization`. */
+    kind: string;
+    /** Free-form field values; consumers map these to the relevant ECHO schema. */
+    fields: Record<string, string>;
+};
+export type ConversationUtterance = {
+    /** Speaker label as written in the fixture (e.g. `A`, `B`, `Rich`). */
+    speaker: string;
+    /** Raw utterance text. */
+    text: string;
+};
+export type ParsedConversation = {
+    entities: ConversationEntity[];
+    utterances: ConversationUtterance[];
+    /** Distinct speaker labels in first-seen order. */
+    speakers: string[];
+};
+/**
+ * Parse the on-disk conversation format into structured data.
+ *
+ * Tolerates leading whitespace inside blocks and ignores blank / `#` comment lines.
+ * Throws if the input doesn't contain a `Conversation:` block — `Entities:` is optional.
+ */
+export declare const parseConversation: (input: string) => ParsedConversation;
+//# sourceMappingURL=conversation-parser.d.ts.map

package/dist/types/src/testing/conversation-parser.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"conversation-parser.d.ts","sourceRoot":"","sources":["../../../../src/testing/conversation-parser.ts"],"names":[],"mappings":"AAQA,MAAM,MAAM,kBAAkB,GAAG;IAC/B,0EAA0E;IAC1E,IAAI,EAAE,MAAM,CAAC;IACb,+EAA+E;IAC/E,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;CAChC,CAAC;AAEF,MAAM,MAAM,qBAAqB,GAAG;IAClC,uEAAuE;IACvE,OAAO,EAAE,MAAM,CAAC;IAChB,0BAA0B;IAC1B,IAAI,EAAE,MAAM,CAAC;CACd,CAAC;AAEF,MAAM,MAAM,kBAAkB,GAAG;IAC/B,QAAQ,EAAE,kBAAkB,EAAE,CAAC;IAC/B,UAAU,EAAE,qBAAqB,EAAE,CAAC;IACpC,mDAAmD;IACnD,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB,CAAC;AAEF;;;;;GAKG;AACH,eAAO,MAAM,iBAAiB,UAAW,MAAM,KAAG,kBAiEjD,CAAC"}

package/dist/types/src/testing/conversation.test.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=conversation.test.d.ts.map

package/dist/types/src/testing/conversation.test.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"conversation.test.d.ts","sourceRoot":"","sources":["../../../../src/testing/conversation.test.ts"],"names":[],"mappings":""}

package/dist/types/src/testing/fixtures.d.ts

@@ -0,0 +1,6 @@
+import { type ScriptedBlock } from './scripted-source';
+/** A short multi-utterance meeting transcript (raw, lower-cased — as Whisper tends to emit). */
+export declare const SAMPLE_MEETING: readonly ScriptedBlock[];
+/** A single-speaker dictation transcript (for the Notes preset). */
+export declare const SAMPLE_NOTES: readonly ScriptedBlock[];
+//# sourceMappingURL=fixtures.d.ts.map

package/dist/types/src/testing/fixtures.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"fixtures.d.ts","sourceRoot":"","sources":["../../../../src/testing/fixtures.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,KAAK,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAEvD,gGAAgG;AAChG,eAAO,MAAM,cAAc,EAAE,SAAS,aAAa,EAIjD,CAAC;AAEH,oEAAoE;AACpE,eAAO,MAAM,YAAY,EAAE,SAAS,aAAa,EAG/C,CAAC"}

package/dist/types/src/testing/index.d.ts

@@ -0,0 +1,5 @@
+export * from './conversation-parser';
+export * from './fixtures';
+export * from './scripted-source';
+export * from './stream-simulator';
+//# sourceMappingURL=index.d.ts.map

package/dist/types/src/testing/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/testing/index.ts"],"names":[],"mappings":"AAIA,cAAc,uBAAuB,CAAC;AACtC,cAAc,YAAY,CAAC;AAC3B,cAAc,mBAAmB,CAAC;AAClC,cAAc,oBAAoB,CAAC"}

package/dist/types/src/testing/scripted-source.d.ts

@@ -0,0 +1,16 @@
+import * as Stream from 'effect/Stream';
+import { TranscriptEvent } from '../TranscriptEvent';
+export type ScriptedBlock = {
+    text: string;
+    started?: string;
+};
+export type ScriptedSourceOptions = {
+    /** Append a trailing silence event (drives on-silence stages). Default: true. */
+    readonly silence?: boolean;
+};
+/**
+ * Build a deterministic {@link TranscriptEvent} stream from a fixture of blocks. Emits a `block`
+ * event per entry, then (by default) a trailing `silence`. No audio or network — for testbench and CI.
+ */
+export declare const scriptedSource: (blocks: readonly ScriptedBlock[], options?: ScriptedSourceOptions) => Stream.Stream<TranscriptEvent>;
+//# sourceMappingURL=scripted-source.d.ts.map

package/dist/types/src/testing/scripted-source.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"scripted-source.d.ts","sourceRoot":"","sources":["../../../../src/testing/scripted-source.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,MAAM,MAAM,eAAe,CAAC;AAExC,OAAO,EAAE,eAAe,EAAE,MAAM,oBAAoB,CAAC;AAErD,MAAM,MAAM,aAAa,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,OAAO,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAE/D,MAAM,MAAM,qBAAqB,GAAG;IAClC,iFAAiF;IACjF,QAAQ,CAAC,OAAO,CAAC,EAAE,OAAO,CAAC;CAC5B,CAAC;AAEF;;;GAGG;AACH,eAAO,MAAM,cAAc,WACjB,SAAS,aAAa,EAAE,YACvB,qBAAqB,KAC7B,MAAM,CAAC,MAAM,CAAC,eAAe,CAY/B,CAAC"}

package/dist/types/src/testing/stream-simulator.d.ts

@@ -0,0 +1,38 @@
+import { type ContentBlock } from '@dxos/types';
+import type { ParsedConversation } from './conversation-parser';
+export type StreamSimulatorOptions = {
+    /**
+     * Average speaking rate, words per second. Used to assign a virtual duration to each
+     * utterance so consecutive utterances land at realistic offsets.
+     * @default 2.5 (≈150 wpm — typical conversational pace)
+     */
+    wordsPerSecond?: number;
+    /**
+     * Gap between consecutive utterances in seconds. Models a brief pause for turn-taking
+     * even when speakers don't actually overlap.
+     * @default 0.5
+     */
+    gapSeconds?: number;
+    /**
+     * Whisper batches typically arrive in chunks of N seconds rather than per-utterance.
+     * When set, utterances shorter than `batchSeconds` are merged into the same batch as
+     * the following utterance from the same speaker. `0` disables merging — every utterance
+     * is its own batch.
+     * @default 0
+     */
+    batchSeconds?: number;
+};
+export type SimulatedBatch = {
+    /** Speaker label from the fixture (`A`, `B`, …). Maps to a virtual `identityDid`. */
+    speaker: string;
+    /** ms offset from the conversation start when this batch is delivered. */
+    offsetMs: number;
+    /** Whisper-shaped transcript blocks (raw text, no enrichment). */
+    blocks: ContentBlock.Transcript[];
+};
+/**
+ * Build the chronological batch schedule for a parsed conversation. The result is a
+ * pure data structure — consumers iterate and drive virtual time themselves.
+ */
+export declare const simulateStream: (conversation: ParsedConversation, options?: StreamSimulatorOptions) => SimulatedBatch[];
+//# sourceMappingURL=stream-simulator.d.ts.map

package/dist/types/src/testing/stream-simulator.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"stream-simulator.d.ts","sourceRoot":"","sources":["../../../../src/testing/stream-simulator.ts"],"names":[],"mappings":"AAaA,OAAO,EAAE,KAAK,YAAY,EAAE,MAAM,aAAa,CAAC;AAEhD,OAAO,KAAK,EAAyB,kBAAkB,EAAE,MAAM,uBAAuB,CAAC;AAEvF,MAAM,MAAM,sBAAsB,GAAG;IACnC;;;;OAIG;IACH,cAAc,CAAC,EAAE,MAAM,CAAC;IAExB;;;;OAIG;IACH,UAAU,CAAC,EAAE,MAAM,CAAC;IAEpB;;;;;;OAMG;IACH,YAAY,CAAC,EAAE,MAAM,CAAC;CACvB,CAAC;AAEF,MAAM,MAAM,cAAc,GAAG;IAC3B,qFAAqF;IACrF,OAAO,EAAE,MAAM,CAAC;IAChB,0EAA0E;IAC1E,QAAQ,EAAE,MAAM,CAAC;IACjB,kEAAkE;IAClE,MAAM,EAAE,YAAY,CAAC,UAAU,EAAE,CAAC;CACnC,CAAC;AAEF;;;GAGG;AACH,eAAO,MAAM,cAAc,iBACX,kBAAkB,YACvB,sBAAsB,KAC9B,cAAc,EAgEhB,CAAC"}