@tscaps/engine 0.1.1 → 0.3.0

package/dist/modules/transcription/WhisperTranscriber.js

@@ -1,10 +1,19 @@
-import { pipeline, env } from '@huggingface/transformers';
 import { Document, Section, Segment, Line, Word, TimeFragment } from '../document/index';
+import { TransformersRuntime } from '../transcription/TransformersRuntime';
+import { WhisperDeviceUnavailableError } from '../transcription/WhisperDeviceUnavailableError';
+import { WhisperInferenceFailedError } from '../transcription/WhisperInferenceFailedError';
+import { WhisperInferenceProgressTracker } from '../transcription/WhisperInferenceProgressTracker';
+import { WhisperWindowCoverage } from '../transcription/WhisperWindowCoverage';
+import { WhisperLoopAbortLogitsProcessor } from '../transcription/WhisperLoopAbortLogitsProcessor';
+import { LogitsProcessorListFactory } from '../transcription/LogitsProcessorListFactory';
+import { WhisperChunkStitcher } from '../transcription/WhisperChunkStitcher';
+import { WhisperModelLoadFailedError } from '../transcription/WhisperModelLoadFailedError';
 export const WHISPER_SAMPLE_RATE = 16_000;
 const MODEL_IDS = {
     'tiny': 'onnx-community/whisper-tiny_timestamped',
     'base': 'onnx-community/whisper-base_timestamped',
     'small': 'onnx-community/whisper-small_timestamped',
+    'medium': 'onnx-community/whisper-medium_timestamped',
     'distil-small.en': 'distil-whisper/distil-small.en',
     'distil-medium.en': 'distil-whisper/distil-medium.en',
 };
@@ -12,6 +21,7 @@ const CHUNK_CONFIG = {
     'tiny': { chunk_length_s: 30, stride_length_s: 5 },
     'base': { chunk_length_s: 30, stride_length_s: 5 },
     'small': { chunk_length_s: 30, stride_length_s: 5 },
+    'medium': { chunk_length_s: 30, stride_length_s: 5 },
     'distil-small.en': { chunk_length_s: 20, stride_length_s: 3 },
     'distil-medium.en': { chunk_length_s: 20, stride_length_s: 3 },
 };
@@ -25,23 +35,54 @@ const CHUNK_CONFIG = {
  * from low-bit weights — bandwidth is the bottleneck there, so q4 gives a
  * 2-4× win.
  *
- * On WASM we run everything at fp32. Counter-intuitively, q4 is SLOWER
- * than fp32 on WASM for Whisper-base-sized models.
+ * On WASM we run q8. On a base-sized model the download drops from 278 MB
+ * (fp32) to 73 MB and end-to-end inference gets ~25% faster than fp32 on
+ * the same isolated worker — halved weight bandwidth beats the int8→fp32
+ * conversion cost by a comfortable margin. Whisper is quantization-
+ * tolerant (WER shift typically well under 1% absolute).
  *
- * INT8 dtypes (`q8`, `int8`, `uint8`) are also unusable on WASM for a
- * separate reason — see huggingface/transformers.js#1635.
- * If #1635 ever closes, q8 on WASM is worth re-benchmarking.
+ * q8 on the `_timestamped` exports crashes session creation under the
+ * `onnxruntime-web` dev build transformers.js@4.2.0 pins — the v4-runtime
+ * optimizer rewrites QDQ into MatMulNBits and demands scale tensors the
+ * older exports do not carry (transformers.js#1707, microsoft/
+ * onnxruntime#28306). The workspace overrides `onnxruntime-web` to 1.27.0,
+ * which carries the fix; the override goes away when transformers.js
+ * ships v4.3.0 with a bundled fix.
+ *
+ * fp16 on WASM crashes a *different* optimizer against the same older
+ * exports (`SimplifiedLayerNormFusion` aborts on a `PrecisionFreeCast`
+ * node name that never got inserted). Not covered by the ORT 1.27 fix
+ * above; deserves its own upstream issue.
+ *
+ * q4 on WASM was measured slower than fp32 with older tooling; not re-
+ * benchmarked and likely irrelevant given q8 already wins on both size
+ * and speed.
  *
  * The distil-whisper repos don't ship the dtypes WebGPU needs
  * (huggingface/transformers.js#1317), so they stay on WASM at q8.
  */
 const DTYPE_BY_DEVICE = {
-    'tiny': { wasm: 'fp32', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
-    'base': { wasm: 'fp32', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
-    'small': { wasm: 'fp32', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
+    'tiny': { wasm: 'q8', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
+    'base': { wasm: 'q8', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
+    'small': { wasm: 'q8', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
+    'medium': { wasm: 'q8', webgpu: { encoder_model: 'fp32', decoder_model_merged: 'q4' } },
     'distil-small.en': { wasm: 'q8', webgpu: null },
     'distil-medium.en': { wasm: 'q8', webgpu: null },
 };
+// Whisper's decoder holds 448 positions, four of which the pipeline spends on
+// the prefix tokens (start-of-transcript, language, task, timestamp mode), so
+// 444 is the model's own generation ceiling. As a cap it is a no-op: it stops
+// generation exactly where the model's `max_length: 448` already would.
+//
+// It is passed for its side effect. transformers.js only enters
+// `_generate_with_seek` — Python's long-form seek loop — when `max_new_tokens`
+// is absent. Layered over `chunk_length_s` windowing (in Python they are
+// alternative long-form strategies, never combined), that loop re-decodes each
+// window's leftover tail zero-padded back to a full 30 s mel, feeding the model
+// long stretches of digital silence — a known hallucination trigger for
+// Whisper. Passing the ceiling as an explicit cap opts out of the layering
+// without changing any decoding limit.
+const MAX_NEW_TOKENS_PER_WINDOW = 444;
 // transformers.js fires both `progress` (per-file) and `progress_total`
 // (aggregate across all files) on the same callback. The per-file event
 // jumps each time a new file starts (it resets to 0), so we listen only
@@ -60,35 +101,150 @@ const AGGREGATE_LOAD_STATUS = 'progress_total';
  * in a Web Worker (where browser audio APIs are unavailable) with a pre-
  * decoded byte-stream decoder.
  *
- * Progress reporting is real-only: the loading stage tracks model weight
- * download from HuggingFace; the inferring stage emits a single boundary
- * event when inference begins, since the underlying pipeline exposes no
- * usable per-chunk hook in v4.
+ * Progress reporting is real-only for both stages. The loading stage
+ * tracks model weight download from HuggingFace. The inferring stage
+ * hooks a `WhisperTextStreamer` into `model.generate` so each
+ * pipeline-window boundary that Whisper predicts translates into an
+ * `[0, 1]` fraction of the audio processed so far. Clips short enough to
+ * fit in a single window emit no boundaries — the field stays at zero
+ * and the consumer can only render that as indeterminate.
+ *
+ * Coverage policy: every pipeline window's decoding is audited. A
+ * non-final window that ended by its own end-of-sequence but left
+ * audio before the next window's start gets one re-decode over real
+ * audio from that point. A window cut from outside — the repetition-
+ * loop guard or the token ceiling — is degenerate: its unclosed tail
+ * is dropped and its region is deliberately left untranscribed, never
+ * retried. A re-decode that degenerates too is discarded the same way.
+ * The final window is reported the same way when cut or stopped
+ * mid-segment; a clean close there is trusted as the end of speech.
  */
 export class WhisperTranscriber {
     decoder;
     model;
     device;
+    chunkStitcher;
+    transformers;
+    processorLists = null;
     pipelinePromise = null;
-    constructor(decoder, { model = 'base', device = 'auto' } = {}) {
+    constructor(decoder, { model = 'base', device = 'auto' } = {}, modelFileCache) {
         this.decoder = decoder;
         this.model = model;
         this.device = device;
-        env.allowLocalModels = false;
+        const { chunk_length_s, stride_length_s } = CHUNK_CONFIG[model];
+        this.chunkStitcher = new WhisperChunkStitcher(chunk_length_s, stride_length_s);
+        this.transformers = new TransformersRuntime(modelFileCache);
     }
     onProgress;
+    onUntranscribedRegion;
     async transcribe(audio, options) {
-        const transcriber = await this.loadPipeline();
+        const transcriber = await this.loadPipelineOrFail();
         const pcm = await this.decoder.decode(audio, WHISPER_SAMPLE_RATE);
-        this.onProgress?.({ stage: 'inferring' });
-        console.time('whisper.inference');
-        const result = (await transcriber(pcm, this.buildPipelineOptions(options)));
-        console.timeEnd('whisper.inference');
-        return this.buildDocument(result.chunks ?? []);
+        this.onProgress?.({ stage: 'inferring', progress: 0 });
+        const durationSeconds = pcm.length / WHISPER_SAMPLE_RATE;
+        const tracker = this.buildProgressTracker(durationSeconds);
+        const coverage = tracker ? new WhisperWindowCoverage() : null;
+        const loopGuard = this.buildLoopGuard(transcriber);
+        const streamer = tracker && coverage
+            ? await this.buildProgressStreamer(transcriber, tracker, coverage, loopGuard)
+            : null;
+        const rawChunks = await this.runInferenceOrFail(transcriber, pcm, options, streamer, loopGuard);
+        const mainChunks = this.chunkStitcher.stitch(rawChunks, durationSeconds);
+        const gaps = tracker && coverage ? this.detectUncoveredGaps(tracker, coverage, durationSeconds) : [];
+        const finalChunks = gaps.length === 0
+            ? mainChunks
+            : await this.redecodeGaps(transcriber, pcm, mainChunks, gaps, options);
+        this.onProgress?.({ stage: 'inferring', progress: 1 });
+        return this.buildDocument(finalChunks);
+    }
+    /**
+     * Builds the progress tracker for this audio, or `null` when a single
+     * pipeline window covers everything — no window boundary is ever
+     * reported then, so the progress bar stays indeterminate.
+     */
+    buildProgressTracker(audioDurationSeconds) {
+        const { chunk_length_s, stride_length_s } = CHUNK_CONFIG[this.model];
+        if (audioDurationSeconds <= chunk_length_s)
+            return null;
+        return new WhisperInferenceProgressTracker(audioDurationSeconds, chunk_length_s, stride_length_s);
+    }
+    /**
+     * Builds a `WhisperTextStreamer` that translates Whisper's per-window
+     * timestamp emissions into `[0, 1]` inference progress and feeds the
+     * per-window coverage record.
+     *
+     * `on_finalize` fires at the end of every `model.generate` call — one
+     * per pipeline window while the long-form layering stays opted out —
+     * which makes it the exact window delimiter the coverage record needs.
+     * It is deliberately not wired to progress: the terminal `progress: 1`
+     * can only come from the caller, after the whole call returns.
+     */
+    async buildProgressStreamer(transcriber, tracker, coverage, loopGuard) {
+        const { WhisperTextStreamer } = await this.transformers.load();
+        const tokenizer = transcriber.tokenizer;
+        return new WhisperTextStreamer(tokenizer, {
+            on_chunk_start: (timeWithinChunk) => coverage.recordTimestamp(timeWithinChunk),
+            on_chunk_end: (timeWithinChunk) => {
+                coverage.recordTimestamp(timeWithinChunk);
+                const progress = tracker.observeChunkEnd(timeWithinChunk);
+                this.onProgress?.({ stage: 'inferring', progress });
+            },
+            on_finalize: () => coverage.windowEnded(loopGuard?.consumeFired() ?? false),
+            token_callback_function: () => coverage.recordToken(),
+            // Silences the default `stdout_write` fallback in TextStreamer,
+            // which prints every decoded token to the console.
+            callback_function: () => { },
+        });
+    }
+    /**
+     * Builds the anti-loop guard for a generation, or `null` when the
+     * model's end-of-sequence token cannot be located — the guard can
+     * only abort by forcing that token.
+     */
+    buildLoopGuard(transcriber) {
+        const model = transcriber.model;
+        const raw = model?.generation_config?.eos_token_id ?? model?.config?.eos_token_id;
+        const endOfSequenceTokenId = Array.isArray(raw) ? raw[0] : raw;
+        if (typeof endOfSequenceTokenId !== 'number') {
+            console.warn('[whisper] end-of-sequence token id not found; running without the loop guard.');
+            return null;
+        }
+        return new WhisperLoopAbortLogitsProcessor(endOfSequenceTokenId);
+    }
+    async loadPipelineOrFail() {
+        try {
+            return await this.loadPipeline();
+        }
+        catch (cause) {
+            if (cause instanceof WhisperDeviceUnavailableError)
+                throw cause;
+            throw new WhisperModelLoadFailedError('Whisper pipeline could not be loaded.', { cause });
+        }
     }
+    async runInferenceOrFail(transcriber, pcm, options, streamer, loopGuard) {
+        try {
+            console.time('whisper.inference');
+            const result = (await transcriber(pcm, await this.buildPipelineOptions(options, streamer, loopGuard)));
+            console.timeEnd('whisper.inference');
+            return result.chunks ?? [];
+        }
+        catch (cause) {
+            throw new WhisperInferenceFailedError('Whisper inference call rejected.', { cause });
+        }
+    }
+    /**
+     * Loads the pipeline once and reuses it for every later call.
+     *
+     * A failed load is *not* kept. The weights arrive over the network
+     * and the common failure is a transfer that dies partway, which the
+     * next attempt may well complete; caching the rejection would make
+     * every retry fail instantly without ever reaching the network
+     * again, leaving a reload as the only way out.
+     */
     loadPipeline() {
         if (this.pipelinePromise === null) {
-            this.pipelinePromise = (async () => {
+            const attempt = (async () => {
+                const { pipeline } = await this.transformers.load();
                 const device = await this.resolveDevice();
                 console.log(`Using device "${device}" for Whisper model "${this.model}".`);
                 const dtype = DTYPE_BY_DEVICE[this.model][device];
@@ -98,6 +254,11 @@ export class WhisperTranscriber {
                     progress_callback: (data) => this.handleLoadProgress(data),
                 });
             })();
+            attempt.catch(() => {
+                if (this.pipelinePromise === attempt)
+                    this.pipelinePromise = null;
+            });
+            this.pipelinePromise = attempt;
         }
         return this.pipelinePromise;
     }
@@ -107,10 +268,10 @@ export class WhisperTranscriber {
             return 'wasm';
         if (this.device === 'webgpu') {
             if (!webgpuSupportedByModel) {
-                throw new Error(`Whisper model "${this.model}" does not support WebGPU; use 'wasm' or 'auto'.`);
+                throw new WhisperDeviceUnavailableError(`Whisper model "${this.model}" does not ship WebGPU artefacts.`);
             }
             if (!(await this.isWebGPUAvailable())) {
-                throw new Error('WebGPU is not available in this browser. Ensure WebGPU is enabled and supported by your browser. Otherwise, switch to CPU in Advanced settings.');
+                throw new WhisperDeviceUnavailableError('WebGPU is not exposed by this browser.');
             }
             return 'webgpu';
         }
@@ -130,15 +291,29 @@ export class WhisperTranscriber {
             return false;
         }
     }
-    buildPipelineOptions(options) {
+    async buildPipelineOptions(options, streamer, loopGuard) {
         return {
             sampling_rate: WHISPER_SAMPLE_RATE,
             return_timestamps: 'word',
+            max_new_tokens: MAX_NEW_TOKENS_PER_WINDOW,
             language: options?.language ?? null,
             task: 'transcribe',
             ...CHUNK_CONFIG[this.model],
+            ...(streamer ? { streamer } : {}),
+            ...(loopGuard ? { logits_processor: await this.buildProcessorList(loopGuard) } : {}),
         };
     }
+    /**
+     * The factory is built from the loaded library, so it is reached
+     * through the same load the pipeline came from rather than held from
+     * construction. By the time any generation runs, that load is settled.
+     */
+    async buildProcessorList(loopGuard) {
+        if (this.processorLists === null) {
+            this.processorLists = new LogitsProcessorListFactory(await this.transformers.load());
+        }
+        return this.processorLists.build(loopGuard);
+    }
     handleLoadProgress(data) {
         if (data.status === AGGREGATE_LOAD_STATUS && data.progress !== undefined) {
             this.onProgress?.({
@@ -147,6 +322,118 @@ export class WhisperTranscriber {
             });
         }
     }
+    /**
+     * The audio regions the main pass left uncovered. Empty when the
+     * number of generation passes does not match the expected window
+     * layout — that means the underlying pipeline changed its long-form
+     * behaviour and per-window attribution can no longer be trusted.
+     */
+    detectUncoveredGaps(tracker, coverage, durationSeconds) {
+        const expectedWindows = tracker.windowLayout.length;
+        if (coverage.windowsObserved !== expectedWindows) {
+            console.warn(`[whisper] expected ${expectedWindows} pipeline windows but observed ` +
+                `${coverage.windowsObserved} generation passes; skipping coverage rescue.`);
+            return [];
+        }
+        const gaps = coverage.uncoveredGaps(tracker.advanceSeconds, MAX_NEW_TOKENS_PER_WINDOW);
+        const finalGap = coverage.finalWindowGap(tracker.advanceSeconds, MAX_NEW_TOKENS_PER_WINDOW, durationSeconds);
+        return finalGap ? [...gaps, finalGap] : gaps;
+    }
+    /**
+     * Replaces the content of every uncovered gap. The main pass's words
+     * inside a gap come from tokens past their window's last closed
+     * segment — the same unclosed tail the reference Whisper decoder
+     * discards before re-decoding — so they are dropped. Dropping is safe
+     * because the word-timestamp alignment is monotonic in token order:
+     * no word of a closed segment lands past that segment's end.
+     *
+     * A rescuable gap gets one extra pass over real audio. A
+     * non-rescuable gap does not — its region is reported and left
+     * untranscribed rather than retried.
+     */
+    async redecodeGaps(transcriber, pcm, main, gaps, options) {
+        const kept = main.filter((chunk) => !this.isInsideAnyGap(chunk, gaps));
+        const rescued = [];
+        for (const gap of gaps) {
+            if (!gap.rescuable) {
+                console.warn(`[whisper] window ended without trustworthy coverage; leaving ` +
+                    `[${gap.startSeconds.toFixed(2)}s, ${gap.endSeconds.toFixed(2)}s] untranscribed.`);
+                this.reportUntranscribed(gap);
+                continue;
+            }
+            rescued.push(...await this.rescueGap(transcriber, pcm, gap, options));
+        }
+        return this.mergeChunks(kept, rescued);
+    }
+    isInsideAnyGap(chunk, gaps) {
+        const anchor = chunk.timestamp[0] ?? chunk.timestamp[1];
+        if (anchor === null)
+            return false;
+        return gaps.some((gap) => anchor >= gap.startSeconds && anchor < gap.endSeconds);
+    }
+    /**
+     * Runs one extra decoding pass over a full window of real audio
+     * starting at the gap and keeps the words that fall inside it. Words
+     * past the gap's end are dropped — the next pipeline window already
+     * covers that region.
+     *
+     * Best-effort by design: the main transcription is already in hand, so
+     * a failed rescue logs and contributes nothing instead of failing the
+     * whole call.
+     */
+    async rescueGap(transcriber, pcm, gap, options) {
+        const { chunk_length_s } = CHUNK_CONFIG[this.model];
+        const label = `[${gap.startSeconds.toFixed(2)}s, ${gap.endSeconds.toFixed(2)}s]`;
+        console.warn(`[whisper] window closed early; re-decoding ${label} from real audio.`);
+        try {
+            const startSample = Math.floor(gap.startSeconds * WHISPER_SAMPLE_RATE);
+            const endSample = Math.min(pcm.length, startSample + chunk_length_s * WHISPER_SAMPLE_RATE);
+            const slice = pcm.subarray(startSample, endSample);
+            const rescueGuard = this.buildLoopGuard(transcriber);
+            const result = (await transcriber(slice, await this.buildPipelineOptions(options, null, rescueGuard)));
+            if (rescueGuard?.consumeFired()) {
+                console.warn(`[whisper] rescue for ${label} degenerated as well; leaving the region untranscribed.`);
+                this.reportUntranscribed(gap);
+                return [];
+            }
+            const recovered = this.chunksWithinGap(result.chunks ?? [], gap);
+            console.log(`[whisper] rescue recovered ${recovered.length} words for ${label}.`);
+            return recovered;
+        }
+        catch (cause) {
+            console.warn('[whisper] gap rescue failed; keeping the main transcription only.', cause);
+            this.reportUntranscribed(gap);
+            return [];
+        }
+    }
+    reportUntranscribed(gap) {
+        this.onUntranscribedRegion?.({ startSeconds: gap.startSeconds, endSeconds: gap.endSeconds });
+    }
+    /**
+     * Shifts slice-relative rescue chunks to absolute time and keeps the
+     * ones anchored inside the gap. A chunk with no usable timestamp
+     * cannot be placed and is dropped.
+     */
+    chunksWithinGap(chunks, gap) {
+        return chunks
+            .map(({ text, timestamp: [start, end] }) => ({
+            text,
+            timestamp: [
+                start === null ? null : start + gap.startSeconds,
+                end === null ? null : end + gap.startSeconds,
+            ],
+        }))
+            .filter(({ timestamp: [start, end] }) => {
+            const anchor = start ?? end;
+            return anchor !== null && anchor < gap.endSeconds;
+        });
+    }
+    mergeChunks(main, rescued) {
+        if (rescued.length === 0)
+            return main;
+        const anchor = ({ timestamp: [start, end] }) => start ?? end ?? 0;
+        return [...main, ...rescued].sort((a, b) => anchor(a) - anchor(b));
+    }
     buildDocument(chunks) {
         const words = chunks
             .map(({ text, timestamp: [start, end] }) => new Word({ text: text.trim(), time: new TimeFragment(start ?? 0, end ?? start ?? 0) }))

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

@@ -1 +1 @@
-{"version":3,"file":"WhisperTranscriber.js","sourceRoot":"","sources":["../../../src/modules/transcription/WhisperTranscriber.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,GAAG,EAAE,MAAM,2BAA2B,CAAC;AAC1D,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAQ/F,MAAM,CAAC,MAAM,mBAAmB,GAAG,MAAM,CAAC;AAgB1C,MAAM,SAAS,GAAiC;IAC9C,MAAM,EAAc,yCAAyC;IAC7D,MAAM,EAAc,yCAAyC;IAC7D,OAAO,EAAa,0CAA0C;IAC9D,iBAAiB,EAAG,gCAAgC;IACpD,kBAAkB,EAAE,iCAAiC;CACtD,CAAC;AAEF,MAAM,YAAY,GAA8E;IAC9F,MAAM,EAAc,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,MAAM,EAAc,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,OAAO,EAAa,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,iBAAiB,EAAG,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,kBAAkB,EAAE,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;CAC/D,CAAC;AAKF;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,eAAe,GAAgE;IACnF,MAAM,EAAc,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACnG,MAAM,EAAc,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACnG,OAAO,EAAa,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACnG,iBAAiB,EAAG,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;IAChD,kBAAkB,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;CACjD,CAAC;AAMF,wEAAwE;AACxE,wEAAwE;AACxE,wEAAwE;AACxE,6EAA6E;AAC7E,MAAM,qBAAqB,GAAG,gBAAgB,CAAC;AAE/C;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,OAAO,kBAAkB;IAMV;IALF,KAAK,CAAe;IACpB,MAAM,CAAgB;IAC/B,eAAe,GAAuC,IAAI,CAAC;IAEnE,YACmB,OAAqB,EACtC,EAAE,KAAK,GAAG,MAAM,EAAE,MAAM,GAAG,MAAM,KAA+B,EAAE;QADjD,YAAO,GAAP,OAAO,CAAc;QAGtC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,GAAG,CAAC,gBAAgB,GAAG,KAAK,CAAC;IAC/B,CAAC;IAED,UAAU,CAA6C;IAEvD,KAAK,CAAC,UAAU,CAAC,KAAW,EAAE,OAA4B;QACxD,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,YAAY,EAAE,CAAC;QAC9C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,KAAK,EAAE,mBAAmB,CAAC,CAAC;QAClE,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,CAAC,CAAC;QAE1C,OAAO,CAAC,IAAI,CAAC,mBAAmB,CAAC,CAAC;QAClC,MAAM,MAAM,GAAG,CAAC,MAAO,WAAgC,CACrD,GAAG,EACH,IAAI,CAAC,oBAAoB,CAAC,OAAO,CAAC,CACnC,CAAkB,CAAC;QACpB,OAAO,CAAC,OAAO,CAAC,mBAAmB,CAAC,CAAC;QAErC,OAAO,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC;IACjD,CAAC;IAEO,YAAY;QAClB,IAAI,IAAI,CAAC,eAAe,KAAK,IAAI,EAAE,CAAC;YAClC,IAAI,CAAC,eAAe,GAAG,CAAC,KAAK,IAAI,EAAE;gBACjC,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,aAAa,EAAE,CAAC;gBAC1C,OAAO,CAAC,GAAG,CAAC,iBAAiB,MAAM,wBAAwB,IAAI,CAAC,KAAK,IAAI,CAAC,CAAC;gBAC3E,MAAM,KAAK,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,CAAE,CAAC;gBACnD,OAAO,QAAQ,CACb,8BAA8B,EAC9B,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,EACrB;oBACE,MAAM;oBACN,KAAK,EAAE,KAAc;oBACrB,iBAAiB,EAAE,CAAC,IAAuB,EAAE,EAAE,CAAC,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC;iBAC9E,CACF,CAAC;YACJ,CAAC,CAAC,EAAE,CAAC;QACP,CAAC;QACD,OAAO,IAAI,CAAC,eAAe,CAAC;IAC9B,CAAC;IAEO,KAAK,CAAC,aAAa;QACzB,MAAM,sBAAsB,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC;QAC3E,IAAI,IAAI,CAAC,MAAM,KAAK,MAAM;YAAE,OAAO,MAAM,CAAC;QAC1C,IAAI,IAAI,CAAC,MAAM,KAAK,QAAQ,EAAE,CAAC;YAC7B,IAAI,CAAC,sBAAsB,EAAE,CAAC;gBAC5B,MAAM,IAAI,KAAK,CACb,kBAAkB,IAAI,CAAC,KAAK,kDAAkD,CAC/E,CAAC;YACJ,CAAC;YACD,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,iBAAiB,EAAE,CAAC,EAAE,CAAC;gBACtC,MAAM,IAAI,KAAK,CACb,iJAAiJ,CAClJ,CAAC;YACJ,CAAC;YACD,OAAO,QAAQ,CAAC;QAClB,CAAC;QACD,IAAI,CAAC,sBAAsB;YAAE,OAAO,MAAM,CAAC;QAC3C,OAAO,CAAC,MAAM,IAAI,CAAC,iBAAiB,EAAE,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IAC9D,CAAC;IAEO,KAAK,CAAC,iBAAiB;QAC7B,IAAI,CAAC;YACH,MAAM,GAAG,GAAI,SAA8D,CAAC,GAAG,CAAC;YAChF,IAAI,CAAC,GAAG;gBAAE,OAAO,KAAK,CAAC;YACvB,MAAM,OAAO,GAAG,MAAM,GAAG,CAAC,cAAc,EAAE,CAAC;YAC3C,OAAO,OAAO,KAAK,IAAI,CAAC;QAC1B,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAEO,oBAAoB,CAAC,OAA4B;QACvD,OAAO;YACL,aAAa,EAAE,mBAAmB;YAClC,iBAAiB,EAAE,MAAe;YAClC,QAAQ,EAAE,OAAO,EAAE,QAAQ,IAAI,IAAI;YACnC,IAAI,EAAE,YAAqB;YAC3B,GAAG,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC;SAC5B,CAAC;IACJ,CAAC;IAEO,kBAAkB,CAAC,IAAuB;QAChD,IAAI,IAAI,CAAC,MAAM,KAAK,qBAAqB,IAAI,IAAI,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;YACzE,IAAI,CAAC,UAAU,EAAE,CAAC;gBAChB,KAAK,EAAE,SAAS;gBAChB,QAAQ,EAAE,IAAI,CAAC,QAAQ,GAAG,GAAG;aAC9B,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAEO,aAAa,CAAC,MAAsB;QAC1C,MAAM,KAAK,GAAG,MAAM;aACjB,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE,EAAE,EAAE,CACzC,IAAI,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,EAAE,IAAI,EAAE,IAAI,YAAY,CAAC,KAAK,IAAI,CAAC,EAAE,GAAG,IAAI,KAAK,IAAI,CAAC,CAAC,EAAE,CAAC,CACvF;aACA,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACpC,OAAO,IAAI,QAAQ,CAAC;YAClB,QAAQ,EAAE,CAAC,IAAI,OAAO,CAAC;oBACrB,QAAQ,EAAE,CAAC,IAAI,OAAO,CAAC,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;oBACzD,IAAI,EAAE,EAAE;iBACT,CAAC,CAAC;SACJ,CAAC,CAAC;IACL,CAAC;CACF"}
+{"version":3,"file":"WhisperTranscriber.js","sourceRoot":"","sources":["../../../src/modules/transcription/WhisperTranscriber.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAC/F,OAAO,EAAE,mBAAmB,EAAE,MAAM,4CAA4C,CAAC;AASjF,OAAO,EAAE,6BAA6B,EAAE,MAAM,sDAAsD,CAAC;AACrG,OAAO,EAAE,2BAA2B,EAAE,MAAM,oDAAoD,CAAC;AACjG,OAAO,EAAE,+BAA+B,EAAE,MAAM,wDAAwD,CAAC;AACzG,OAAO,EAAE,qBAAqB,EAAoB,MAAM,8CAA8C,CAAC;AACvG,OAAO,EAAE,+BAA+B,EAAE,MAAM,wDAAwD,CAAC;AACzG,OAAO,EAAE,0BAA0B,EAAE,MAAM,mDAAmD,CAAC;AAC/F,OAAO,EAAE,oBAAoB,EAAqB,MAAM,6CAA6C,CAAC;AACtG,OAAO,EAAE,2BAA2B,EAAE,MAAM,oDAAoD,CAAC;AAEjG,MAAM,CAAC,MAAM,mBAAmB,GAAG,MAAM,CAAC;AAiB1C,MAAM,SAAS,GAAiC;IAC9C,MAAM,EAAc,yCAAyC;IAC7D,MAAM,EAAc,yCAAyC;IAC7D,OAAO,EAAa,0CAA0C;IAC9D,QAAQ,EAAY,2CAA2C;IAC/D,iBAAiB,EAAG,gCAAgC;IACpD,kBAAkB,EAAE,iCAAiC;CACtD,CAAC;AAEF,MAAM,YAAY,GAA8E;IAC9F,MAAM,EAAc,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,MAAM,EAAc,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,OAAO,EAAa,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,QAAQ,EAAY,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,iBAAiB,EAAG,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;IAC9D,kBAAkB,EAAE,EAAE,cAAc,EAAE,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE;CAC/D,CAAC;AAKF;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmCG;AACH,MAAM,eAAe,GAAgE;IACnF,MAAM,EAAc,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACjG,MAAM,EAAc,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACjG,OAAO,EAAa,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACjG,QAAQ,EAAY,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,EAAE;IACjG,iBAAiB,EAAG,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;IAChD,kBAAkB,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;CACjD,CAAC;AAEF,8EAA8E;AAC9E,8EAA8E;AAC9E,8EAA8E;AAC9E,wEAAwE;AACxE,EAAE;AACF,gEAAgE;AAChE,+EAA+E;AAC/E,yEAAyE;AACzE,+EAA+E;AAC/E,gFAAgF;AAChF,wEAAwE;AACxE,2EAA2E;AAC3E,uCAAuC;AACvC,MAAM,yBAAyB,GAAG,GAAG,CAAC;AAKtC,wEAAwE;AACxE,wEAAwE;AACxE,wEAAwE;AACxE,6EAA6E;AAC7E,MAAM,qBAAqB,GAAG,gBAAgB,CAAC;AAE/C;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AACH,MAAM,OAAO,kBAAkB;IASV;IARF,KAAK,CAAe;IACpB,MAAM,CAAgB;IACtB,aAAa,CAAuB;IACpC,YAAY,CAAsB;IAC3C,cAAc,GAAsC,IAAI,CAAC;IACzD,eAAe,GAAuC,IAAI,CAAC;IAEnE,YACmB,OAAqB,EACtC,EAAE,KAAK,GAAG,MAAM,EAAE,MAAM,GAAG,MAAM,KAA+B,EAAE,EAClE,cAA+B;QAFd,YAAO,GAAP,OAAO,CAAc;QAItC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,MAAM,EAAE,cAAc,EAAE,eAAe,EAAE,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC;QAChE,IAAI,CAAC,aAAa,GAAG,IAAI,oBAAoB,CAAC,cAAc,EAAE,eAAe,CAAC,CAAC;QAC/E,IAAI,CAAC,YAAY,GAAG,IAAI,mBAAmB,CAAC,cAAc,CAAC,CAAC;IAC9D,CAAC;IAED,UAAU,CAA6C;IACvD,qBAAqB,CAAyC;IAE9D,KAAK,CAAC,UAAU,CAAC,KAAW,EAAE,OAA4B;QACxD,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,kBAAkB,EAAE,CAAC;QACpD,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,KAAK,EAAE,mBAAmB,CAAC,CAAC;QAClE,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QACvD,MAAM,eAAe,GAAG,GAAG,CAAC,MAAM,GAAG,mBAAmB,CAAC;QACzD,MAAM,OAAO,GAAG,IAAI,CAAC,oBAAoB,CAAC,eAAe,CAAC,CAAC;QAC3D,MAAM,QAAQ,GAAG,OAAO,CAAC,CAAC,CAAC,IAAI,qBAAqB,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;QAC9D,MAAM,SAAS,GAAG,IAAI,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC;QACnD,MAAM,QAAQ,GAAG,OAAO,IAAI,QAAQ;YAClC,CAAC,CAAC,MAAM,IAAI,CAAC,qBAAqB,CAAC,WAAW,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,CAAC;YAC7E,CAAC,CAAC,IAAI,CAAC;QACT,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,kBAAkB,CAAC,WAAW,EAAE,GAAG,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,CAAC,CAAC;QAChG,MAAM,UAAU,GAAG,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,SAAS,EAAE,eAAe,CAAC,CAAC;QACzE,MAAM,IAAI,GAAG,OAAO,IAAI,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,mBAAmB,CAAC,OAAO,EAAE,QAAQ,EAAE,eAAe,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QACrG,MAAM,WAAW,GAAG,IAAI,CAAC,MAAM,KAAK,CAAC;YACnC,CAAC,CAAC,UAAU;YACZ,CAAC,CAAC,MAAM,IAAI,CAAC,YAAY,CAAC,WAAW,EAAE,GAAG,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC;QACzE,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QACvD,OAAO,IAAI,CAAC,aAAa,CAAC,WAAW,CAAC,CAAC;IACzC,CAAC;IAED;;;;OAIG;IACK,oBAAoB,CAAC,oBAA4B;QACvD,MAAM,EAAE,cAAc,EAAE,eAAe,EAAE,GAAG,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACrE,IAAI,oBAAoB,IAAI,cAAc;YAAE,OAAO,IAAI,CAAC;QACxD,OAAO,IAAI,+BAA+B,CACxC,oBAAoB,EACpB,cAAc,EACd,eAAe,CAChB,CAAC;IACJ,CAAC;IAED;;;;;;;;;;OAUG;IACK,KAAK,CAAC,qBAAqB,CACjC,WAAiD,EACjD,OAAwC,EACxC,QAA+B,EAC/B,SAAiD;QAEjD,MAAM,EAAE,mBAAmB,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;QAC/D,MAAM,SAAS,GAAI,WAA+C,CAAC,SAAS,CAAC;QAC7E,OAAO,IAAI,mBAAmB,CAAC,SAAS,EAAE;YACxC,cAAc,EAAE,CAAC,eAAuB,EAAE,EAAE,CAAC,QAAQ,CAAC,eAAe,CAAC,eAAe,CAAC;YACtF,YAAY,EAAE,CAAC,eAAuB,EAAE,EAAE;gBACxC,QAAQ,CAAC,eAAe,CAAC,eAAe,CAAC,CAAC;gBAC1C,MAAM,QAAQ,GAAG,OAAO,CAAC,eAAe,CAAC,eAAe,CAAC,CAAC;gBAC1D,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,CAAC;YACtD,CAAC;YACD,WAAW,EAAE,GAAG,EAAE,CAAC,QAAQ,CAAC,WAAW,CAAC,SAAS,EAAE,YAAY,EAAE,IAAI,KAAK,CAAC;YAC3E,uBAAuB,EAAE,GAAG,EAAE,CAAC,QAAQ,CAAC,WAAW,EAAE;YACrD,gEAAgE;YAChE,mDAAmD;YACnD,iBAAiB,EAAE,GAAG,EAAE,GAAE,CAAC;SAC5B,CAAC,CAAC;IACL,CAAC;IAED;;;;OAIG;IACK,cAAc,CACpB,WAAiD;QAEjD,MAAM,KAAK,GAAI,WAEb,CAAC,KAAK,CAAC;QACT,MAAM,GAAG,GAAG,KAAK,EAAE,iBAAiB,EAAE,YAAY,IAAI,KAAK,EAAE,MAAM,EAAE,YAAY,CAAC;QAClF,MAAM,oBAAoB,GAAG,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;QAC/D,IAAI,OAAO,oBAAoB,KAAK,QAAQ,EAAE,CAAC;YAC7C,OAAO,CAAC,IAAI,CAAC,+EAA+E,CAAC,CAAC;YAC9F,OAAO,IAAI,CAAC;QACd,CAAC;QACD,OAAO,IAAI,+BAA+B,CAAC,oBAAoB,CAAC,CAAC;IACnE,CAAC;IAEO,KAAK,CAAC,kBAAkB;QAC9B,IAAI,CAAC;YACH,OAAO,MAAM,IAAI,CAAC,YAAY,EAAE,CAAC;QACnC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,IAAI,KAAK,YAAY,6BAA6B;gBAAE,MAAM,KAAK,CAAC;YAChE,MAAM,IAAI,2BAA2B,CAAC,uCAAuC,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;QAC5F,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,kBAAkB,CAC9B,WAAiD,EACjD,GAAiB,EACjB,OAA4B,EAC5B,QAAqC,EACrC,SAAkD;QAElD,IAAI,CAAC;YACH,OAAO,CAAC,IAAI,CAAC,mBAAmB,CAAC,CAAC;YAClC,MAAM,MAAM,GAAG,CAAC,MAAO,WAAgC,CACrD,GAAG,EACH,MAAM,IAAI,CAAC,oBAAoB,CAAC,OAAO,EAAE,QAAQ,EAAE,SAAS,CAAC,CAC9D,CAAkB,CAAC;YACpB,OAAO,CAAC,OAAO,CAAC,mBAAmB,CAAC,CAAC;YACrC,OAAO,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC;QAC7B,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,IAAI,2BAA2B,CAAC,kCAAkC,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;QACvF,CAAC;IACH,CAAC;IAED;;;;;;;;OAQG;IACK,YAAY;QAClB,IAAI,IAAI,CAAC,eAAe,KAAK,IAAI,EAAE,CAAC;YAClC,MAAM,OAAO,GAAG,CAAC,KAAK,IAAI,EAAE;gBAC1B,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;gBACpD,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,aAAa,EAAE,CAAC;gBAC1C,OAAO,CAAC,GAAG,CAAC,iBAAiB,MAAM,wBAAwB,IAAI,CAAC,KAAK,IAAI,CAAC,CAAC;gBAC3E,MAAM,KAAK,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,CAAE,CAAC;gBACnD,OAAO,QAAQ,CACb,8BAA8B,EAC9B,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,EACrB;oBACE,MAAM;oBACN,KAAK,EAAE,KAAc;oBACrB,iBAAiB,EAAE,CAAC,IAAuB,EAAE,EAAE,CAAC,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC;iBAC9E,CACF,CAAC;YACJ,CAAC,CAAC,EAAE,CAAC;YACL,OAAO,CAAC,KAAK,CAAC,GAAG,EAAE;gBACjB,IAAI,IAAI,CAAC,eAAe,KAAK,OAAO;oBAAE,IAAI,CAAC,eAAe,GAAG,IAAI,CAAC;YACpE,CAAC,CAAC,CAAC;YACH,IAAI,CAAC,eAAe,GAAG,OAAO,CAAC;QACjC,CAAC;QACD,OAAO,IAAI,CAAC,eAAe,CAAC;IAC9B,CAAC;IAEO,KAAK,CAAC,aAAa;QACzB,MAAM,sBAAsB,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC;QAC3E,IAAI,IAAI,CAAC,MAAM,KAAK,MAAM;YAAE,OAAO,MAAM,CAAC;QAC1C,IAAI,IAAI,CAAC,MAAM,KAAK,QAAQ,EAAE,CAAC;YAC7B,IAAI,CAAC,sBAAsB,EAAE,CAAC;gBAC5B,MAAM,IAAI,6BAA6B,CACrC,kBAAkB,IAAI,CAAC,KAAK,mCAAmC,CAChE,CAAC;YACJ,CAAC;YACD,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,iBAAiB,EAAE,CAAC,EAAE,CAAC;gBACtC,MAAM,IAAI,6BAA6B,CACrC,wCAAwC,CACzC,CAAC;YACJ,CAAC;YACD,OAAO,QAAQ,CAAC;QAClB,CAAC;QACD,IAAI,CAAC,sBAAsB;YAAE,OAAO,MAAM,CAAC;QAC3C,OAAO,CAAC,MAAM,IAAI,CAAC,iBAAiB,EAAE,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;IAC9D,CAAC;IAEO,KAAK,CAAC,iBAAiB;QAC7B,IAAI,CAAC;YACH,MAAM,GAAG,GAAI,SAA8D,CAAC,GAAG,CAAC;YAChF,IAAI,CAAC,GAAG;gBAAE,OAAO,KAAK,CAAC;YACvB,MAAM,OAAO,GAAG,MAAM,GAAG,CAAC,cAAc,EAAE,CAAC;YAC3C,OAAO,OAAO,KAAK,IAAI,CAAC;QAC1B,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,oBAAoB,CAChC,OAA4B,EAC5B,QAAqC,EACrC,SAAkD;QAElD,OAAO;YACL,aAAa,EAAE,mBAAmB;YAClC,iBAAiB,EAAE,MAAe;YAClC,cAAc,EAAE,yBAAyB;YACzC,QAAQ,EAAE,OAAO,EAAE,QAAQ,IAAI,IAAI;YACnC,IAAI,EAAE,YAAqB;YAC3B,GAAG,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC;YAC3B,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YACjC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,MAAM,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SACrF,CAAC;IACJ,CAAC;IAED;;;;OAIG;IACK,KAAK,CAAC,kBAAkB,CAC9B,SAA0C;QAE1C,IAAI,IAAI,CAAC,cAAc,KAAK,IAAI,EAAE,CAAC;YACjC,IAAI,CAAC,cAAc,GAAG,IAAI,0BAA0B,CAAC,MAAM,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,CAAC;QACvF,CAAC;QACD,OAAO,IAAI,CAAC,cAAc,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;IAC9C,CAAC;IAEO,kBAAkB,CAAC,IAAuB;QAChD,IAAI,IAAI,CAAC,MAAM,KAAK,qBAAqB,IAAI,IAAI,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;YACzE,IAAI,CAAC,UAAU,EAAE,CAAC;gBAChB,KAAK,EAAE,SAAS;gBAChB,QAAQ,EAAE,IAAI,CAAC,QAAQ,GAAG,GAAG;aAC9B,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED;;;;;OAKG;IACK,mBAAmB,CACzB,OAAwC,EACxC,QAA+B,EAC/B,eAAuB;QAEvB,MAAM,eAAe,GAAG,OAAO,CAAC,YAAY,CAAC,MAAM,CAAC;QACpD,IAAI,QAAQ,CAAC,eAAe,KAAK,eAAe,EAAE,CAAC;YACjD,OAAO,CAAC,IAAI,CACV,sBAAsB,eAAe,iCAAiC;gBACtE,GAAG,QAAQ,CAAC,eAAe,+CAA+C,CAC3E,CAAC;YACF,OAAO,EAAE,CAAC;QACZ,CAAC;QACD,MAAM,IAAI,GAAG,QAAQ,CAAC,aAAa,CAAC,OAAO,CAAC,cAAc,EAAE,yBAAyB,CAAC,CAAC;QACvF,MAAM,QAAQ,GAAG,QAAQ,CAAC,cAAc,CAAC,OAAO,CAAC,cAAc,EAAE,yBAAyB,EAAE,eAAe,CAAC,CAAC;QAC7G,OAAO,QAAQ,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC/C,CAAC;IAED;;;;;;;;;;;OAWG;IACK,KAAK,CAAC,YAAY,CACxB,WAAiD,EACjD,GAAiB,EACjB,IAAoB,EACpB,IAAmB,EACnB,OAA4B;QAE5B,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,CAAC;QACvE,MAAM,OAAO,GAAmB,EAAE,CAAC;QACnC,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACvB,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC;gBACnB,OAAO,CAAC,IAAI,CACV,+DAA+D;oBAC/D,IAAI,GAAG,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,mBAAmB,CAClF,CAAC;gBACF,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,CAAC;gBAC9B,SAAS;YACX,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,GAAG,MAAM,IAAI,CAAC,SAAS,CAAC,WAAW,EAAE,GAAG,EAAE,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC;QACxE,CAAC;QACD,OAAO,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IACzC,CAAC;IAEO,cAAc,CAAC,KAAmB,EAAE,IAAmB;QAC7D,MAAM,MAAM,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;QACxD,IAAI,MAAM,KAAK,IAAI;YAAE,OAAO,KAAK,CAAC;QAClC,OAAO,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,MAAM,IAAI,GAAG,CAAC,YAAY,IAAI,MAAM,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC;IACnF,CAAC;IAED;;;;;;;;;OASG;IACK,KAAK,CAAC,SAAS,CACrB,WAAiD,EACjD,GAAiB,EACjB,GAAgB,EAChB,OAA4B;QAE5B,MAAM,EAAE,cAAc,EAAE,GAAG,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACpD,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC;QACjF,OAAO,CAAC,IAAI,CAAC,8CAA8C,KAAK,mBAAmB,CAAC,CAAC;QACrF,IAAI,CAAC;YACH,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,YAAY,GAAG,mBAAmB,CAAC,CAAC;YACvE,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,EAAE,WAAW,GAAG,cAAc,GAAG,mBAAmB,CAAC,CAAC;YAC3F,MAAM,KAAK,GAAG,GAAG,CAAC,QAAQ,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACnD,MAAM,WAAW,GAAG,IAAI,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC;YACrD,MAAM,MAAM,GAAG,CAAC,MAAO,WAAgC,CACrD,KAAK,EACL,MAAM,IAAI,CAAC,oBAAoB,CAAC,OAAO,EAAE,IAAI,EAAE,WAAW,CAAC,CAC5D,CAAkB,CAAC;YACpB,IAAI,WAAW,EAAE,YAAY,EAAE,EAAE,CAAC;gBAChC,OAAO,CAAC,IAAI,CAAC,wBAAwB,KAAK,yDAAyD,CAAC,CAAC;gBACrG,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,CAAC;gBAC9B,OAAO,EAAE,CAAC;YACZ,CAAC;YACD,MAAM,SAAS,GAAG,IAAI,CAAC,eAAe,CAAC,MAAM,CAAC,MAAM,IAAI,EAAE,EAAE,GAAG,CAAC,CAAC;YACjE,OAAO,CAAC,GAAG,CAAC,8BAA8B,SAAS,CAAC,MAAM,cAAc,KAAK,GAAG,CAAC,CAAC;YAClF,OAAO,SAAS,CAAC;QACnB,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,OAAO,CAAC,IAAI,CAAC,mEAAmE,EAAE,KAAK,CAAC,CAAC;YACzF,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,CAAC;YAC9B,OAAO,EAAE,CAAC;QACZ,CAAC;IACH,CAAC;IAEO,mBAAmB,CAAC,GAAgB;QAC1C,IAAI,CAAC,qBAAqB,EAAE,CAAC,EAAE,YAAY,EAAE,GAAG,CAAC,YAAY,EAAE,UAAU,EAAE,GAAG,CAAC,UAAU,EAAE,CAAC,CAAC;IAC/F,CAAC;IAED;;;;OAIG;IACK,eAAe,CAAC,MAAsB,EAAE,GAAgB;QAC9D,OAAO,MAAM;aACV,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE,EAAgB,EAAE,CAAC,CAAC;YACzD,IAAI;YACJ,SAAS,EAAE;gBACT,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,GAAG,CAAC,YAAY;gBAChD,GAAG,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC,YAAY;aAC7C;SACF,CAAC,CAAC;aACF,MAAM,CAAC,CAAC,EAAE,SAAS,EAAE,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE,EAAE,EAAE;YACtC,MAAM,MAAM,GAAG,KAAK,IAAI,GAAG,CAAC;YAC5B,OAAO,MAAM,KAAK,IAAI,IAAI,MAAM,GAAG,GAAG,CAAC,UAAU,CAAC;QACpD,CAAC,CAAC,CAAC;IACP,CAAC;IAEO,WAAW,CAAC,IAAoB,EAAE,OAAuB;QAC/D,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC;QACtC,MAAM,MAAM,GAAG,CAAC,EAAE,SAAS,EAAE,CAAC,KAAK,EAAE,GAAG,CAAC,EAAgB,EAAE,EAAE,CAAC,KAAK,IAAI,GAAG,IAAI,CAAC,CAAC;QAChF,OAAO,CAAC,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;IACrE,CAAC;IAEO,aAAa,CAAC,MAAsB;QAC1C,MAAM,KAAK,GAAG,MAAM;aACjB,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE,EAAE,EAAE,CACzC,IAAI,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,EAAE,IAAI,EAAE,IAAI,YAAY,CAAC,KAAK,IAAI,CAAC,EAAE,GAAG,IAAI,KAAK,IAAI,CAAC,CAAC,EAAE,CAAC,CACvF;aACA,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACpC,OAAO,IAAI,QAAQ,CAAC;YAClB,QAAQ,EAAE,CAAC,IAAI,OAAO,CAAC;oBACrB,QAAQ,EAAE,CAAC,IAAI,OAAO,CAAC,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;oBACzD,IAAI,EAAE,EAAE;iBACT,CAAC,CAAC;SACJ,CAAC,CAAC;IACL,CAAC;CACF"}

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

@@ -0,0 +1,79 @@
+/** An audio region, in absolute seconds, that no pipeline window transcribed. */
+export interface CoverageGap {
+    startSeconds: number;
+    endSeconds: number;
+    /**
+     * Whether one re-decode over real audio may recover the region.
+     * False when the window's output was cut from outside — loop guard
+     * or token ceiling — because retrying reproduces the failure, and
+     * for final-window gaps, where re-decoding trailing audio risks
+     * hallucinating words over silence.
+     */
+    rescuable: boolean;
+}
+/**
+ * Records what the decoder did in each pipeline window — how far its
+ * last real segment got, how many tokens it emitted, whether the loop
+ * guard cut it, whether it stopped mid-segment — and answers which
+ * audio regions were left uncovered.
+ *
+ * Segment boundaries are counted with window-local parity: within one
+ * window the decoder strictly alternates open / close starting at
+ * open, so odd-numbered events open a segment and even-numbered ones
+ * close it. The upstream streamer's own open/close labels are not
+ * trusted — its alternation state carries across windows, so one
+ * window that stops mid-segment mislabels every event that follows.
+ *
+ * Only segments whose close timestamp is strictly greater than their
+ * open timestamp count as "closed". A degenerate repetition loop can
+ * emit `<|t|>text<|t|>` pairs at a single timestamp forever — trusting
+ * those bogus closes leaves the whole loop tail looking legitimate
+ * and defeats every downstream cutoff.
+ */
+export declare class WhisperWindowCoverage {
+    private readonly windows;
+    private currentLastClosedEnd;
+    private currentTokenCount;
+    private currentTimestampEvents;
+    private currentOpenTimestamp;
+    /** Feeds one segment-boundary timestamp, relative to the current window. */
+    recordTimestamp(timeWithinWindow: number): void;
+    private isOpenEvent;
+    private isRealClose;
+    /** Counts one generated token towards the current window. */
+    recordToken(): void;
+    /** Marks the end of the current window's generation pass. */
+    windowEnded(abortedByLoopGuard: boolean): void;
+    /** Number of generation passes observed so far. */
+    get windowsObserved(): number;
+    /**
+     * The uncovered regions among non-final windows. A window whose
+     * trustworthy coverage ends before the pipeline's window advance
+     * leaves the audio from that point up to the next window's start
+     * transcribed by nobody.
+     *
+     * A window is degenerate when the loop guard cut it or its token
+     * count reached `tokenCeiling` — either way it did not end by its
+     * own end-of-sequence, and its gap is not rescuable.
+     */
+    uncoveredGaps(advanceSeconds: number, tokenCeiling: number): CoverageGap[];
+    /**
+     * The uncovered tail of the final window, or `null` when there is
+     * nothing trustworthy to report. Reported only when that window's
+     * generation was cut from outside or stopped mid-segment. A clean
+     * close is trusted as the genuine end of speech: with no following
+     * window, an early-but-clean close cannot be told apart from a video
+     * that simply ends in silence. Never rescuable.
+     */
+    finalWindowGap(advanceSeconds: number, tokenCeiling: number, audioEndSeconds: number): CoverageGap | null;
+    private isDegenerate;
+    /**
+     * Where the window's trustworthy coverage ends. A clean window that
+     * closed no segment reports `null`: re-decoding its identical input
+     * is deterministic and would reproduce the same empty result. A
+     * degenerate window with no closed segment still gaps from zero —
+     * its region gets dropped, not re-decoded.
+     */
+    private gapStartFor;
+}
+//# sourceMappingURL=WhisperWindowCoverage.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"WhisperWindowCoverage.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/WhisperWindowCoverage.ts"],"names":[],"mappings":"AAAA,iFAAiF;AACjF,MAAM,WAAW,WAAW;IAC1B,YAAY,EAAE,MAAM,CAAC;IACrB,UAAU,EAAE,MAAM,CAAC;IACnB;;;;;;OAMG;IACH,SAAS,EAAE,OAAO,CAAC;CACpB;AASD;;;;;;;;;;;;;;;;;;GAkBG;AACH,qBAAa,qBAAqB;IAChC,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAsB;IAC9C,OAAO,CAAC,oBAAoB,CAAuB;IACnD,OAAO,CAAC,iBAAiB,CAAK;IAC9B,OAAO,CAAC,sBAAsB,CAAK;IACnC,OAAO,CAAC,oBAAoB,CAAuB;IAEnD,4EAA4E;IAC5E,eAAe,CAAC,gBAAgB,EAAE,MAAM,GAAG,IAAI;IAY/C,OAAO,CAAC,WAAW;IAInB,OAAO,CAAC,WAAW;IAInB,6DAA6D;IAC7D,WAAW,IAAI,IAAI;IAInB,6DAA6D;IAC7D,WAAW,CAAC,kBAAkB,EAAE,OAAO,GAAG,IAAI;IAa9C,mDAAmD;IACnD,IAAI,eAAe,IAAI,MAAM,CAE5B;IAED;;;;;;;;;OASG;IACH,aAAa,CAAC,cAAc,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,GAAG,WAAW,EAAE;IAiB1E;;;;;;;OAOG;IACH,cAAc,CAAC,cAAc,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,eAAe,EAAE,MAAM,GAAG,WAAW,GAAG,IAAI;IAUzG,OAAO,CAAC,YAAY;IAIpB;;;;;;OAMG;IACH,OAAO,CAAC,WAAW;CAIpB"}