- New buildCacheKey function creates unique identifiers for TTS cache entries.
- Cache keys now include sentence, voice, speed, provider, and model parameters.
- Prevents serving cached audio or alignment data that mismatches current TTS parameters.
- Removes redundant audioCache.clear() calls when voice or speed change, as entries are now distinct.
Abstracted direct fetch calls across components and contexts into new functions within `src/lib/client.ts`. This provides a consistent and centralized interface for interacting with backend APIs.
- Introduced `src/lib/client.ts` to encapsulate API request logic.
- Standardized audio buffer types (`TTSAudioBuffer`, `TTSAudioBytes`) in `src/types/tts.ts`.
- Moved client-specific request types (`TTSRequestPayload`, `TTSRequestHeaders`, `TTSRetryOptions`) to `src/types/client.ts`.
- Updated API routes and consumer components/contexts to leverage the new client library functions and type definitions.
- Removed `src/utils/audio.ts` as its utility functions are now part of `src/lib/client.ts`.
- Relocated all audiobook-related API routes from `/api/audio/convert/*` to `/api/audiobook/*`.
- This change affects endpoints for chapter conversion, retrieval, deletion, and overall audiobook status.
- Updated client-side calls in `AudiobookExportModal.tsx`, `EPUBContext.tsx`, and `PDFContext.tsx` to reflect the new API paths.
- Modified API tests (`api.spec.ts`, `export.spec.ts`) to target the restructured endpoints.
- The new API structure provides better organization and a clearer, more consistent interface for audiobook functionality.
- Introduce `/api/whisper` endpoint which uses `whisper.cpp` (via a `WHISPER_CPP_BIN` executable) and `ffmpeg` to generate word-level audio alignments from provided audio and text.
- Integrate word-level alignments into `TTSContext`, tracking the currently spoken word based on audio seek position and provided timestamps. Alignments are cached in-memory and fetched asynchronously.
- Add new configuration options (`pdfWordHighlightEnabled`, `epubWordHighlightEnabled`) to `ConfigContext` and `Dexie` for enabling/disabling the feature.
- Implement visual word highlighting in both `PDFViewer` and `EPUBViewer` by mapping TTS-aligned words to rendered text elements.
- Enhance `EPUBContext` and `PDFContext` with new `highlightWordIndex` and `clearWordHighlights` functions, utilizing fuzzy string matching (`cmpstr`) to robustly align spoken words with displayed text for accurate highlighting.
- Update `DocumentSettings` to include user-facing toggles for the new highlighting modes.
- Removed `onProgress` callback from `regenerateChapter` and related functions across `TTS`, `EPUB`, and `PDF` contexts to simplify the chapter regeneration API.
- Updated `AudiobookExportModal` to align with the refined regeneration API, including removing granular chapter progress display and adding a hint about TTS caching behavior.
- Introduced `TTSAudiobookChapter` interface and renamed `ContinuationMergeResult` to `TTSSmartMergeResult` and `PageTurnEstimate` to `TTSPageTurnEstimate` for better type consistency and clarity.
- Applied minor styling adjustments to buttons and listbox components in modals for visual consistency.
- Added `'use client'` directive to several client-side components for Next.js 13+ compatibility.
- Updated Dockerfile build command from `pnpm run build` to `pnpm build`.
- Added Playwright tests to verify backend chapter state after regeneration.
- Reorganized utility modules from `src/utils` to `src/lib` for clearer separation of concerns.
- Introduced new, dedicated type definitions in `src/types` for improved type safety in configuration and TTS API interactions.
- Replaced `src/types/appConfig.ts` with `src/types/config.ts`.
- Added `src/types/tts.ts` for TTS request payloads, error structures, and retry options.
- Updated module imports across several contexts (`Config`, `EPUB`, `HTML`, `PDF`, `TTS`) and components to reflect the new `lib` and `types` locations.
- Enhanced TTS API request and error handling in `src/app/api/tts/route.ts` and TTS-consuming contexts with explicit types.
- Simplified `ProgressCard`, `ProgressPopup`, and `AudiobookExportModal` components by removing the `isProcessing` prop, centralizing processing state management.
- Streamlined `HTMLContext` by removing `createFullAudioBook` and `isAudioCombining` properties, focusing its scope.
Migrates scattered configuration items and last document locations
from the legacy `config` table to new, dedicated Dexie stores.
- Introduces a singleton `app-config` table (`AppConfigRow`) to standardize
and structure all application settings in a single object.
- Creates a `last-locations` table for efficient storage of document read positions.
- Updates `ConfigContext` to utilize `dexie-react-hooks`' `useLiveQuery`
for reactive and simplified state management.
- Implements a database upgrade path (from DB_VERSION 4 to 5) to migrate
all existing user settings seamlessly.
- Simplifies config access and updates with new `getAppConfig` and `updateAppConfig`
utility functions.
- Removes the deprecated `config` table after successful migration.
- Updates consumers like `SettingsModal` and test helpers to align with
the new config structure.
- Offload computationally intensive text matching for real-time highlighting to a dedicated Web Worker, ensuring the main thread remains responsive during playback.
- Implement a new overlay-based highlighting system that renders independently of the PDF's text layer, providing smoother and more reliable visual feedback without interfering with document rendering.
- Introduce a new setting allowing users to enable or disable real-time text highlighting in PDFs, offering personalized control over the reading interface.
- Upgrade the underlying text comparison algorithm to Dice similarity for more accurate and context-aware matching of spoken words to on-screen text, improving synchronization precision.
- Improve sentence boundary detection, especially for quoted dialogue and complex structures, by enhancing the NLP processing logic, leading to a more natural audio-text flow.
Replaces custom IndexedDB implementation with Dexie ORM, eliminating 850+ lines of
boilerplate code and introducing reactive live queries across all document types.
Transforms document management from imperative refresh patterns to automatic
reactive updates using dexie-react-hooks.
Simplifies TTS backend by removing concurrency semaphore while maintaining
request de-duplication through in-flight tracking. Streamlines document hooks
by removing manual state management and refresh methods.
Updates package dependencies and type definitions to support new database
architecture while maintaining full backward compatibility for existing
documents and settings.
BREAKING CHANGE: Document hooks no longer expose refresh() methods as updates
are now reactive through live queries.
introduce configurable smart sentence splitting with persisted state,
UI toggle, and EPUB/PDF contexts that send continuation metadata.
enhance TTS pipeline to merge cross-page sentences, manage carryover,
and trigger visual page changes during playback for smoother narration.
update README with kokoro quick-start guidance, remove the legacy
issues mapping doc, and add deterministic TTS mocks plus sample audio
for Playwright tests.
- Introduce voice utils (model detection, voice parsing/weights, limits)
- Enable Kokoro multi-voice strings across OpenAI/Deepinfra/custom providers
- Normalize non-Kokoro voices to single token for SDK calls
- Expose full Kokoro voice list for custom-openai Kokoro models
- Update TTS API to return audio as ArrayBuffer and improve logging
- Add multi-select UI for Kokoro voices with provider-based clamping
- Preserve Kokoro voice strings in TTSContext and coalesce restarts
- Merge multi-sentence quoted dialogue in NLP sentence splitter
- Update tests for single and multi-voice selection flows
Add comprehensive multi-provider TTS support enabling users to choose between OpenAI, Deepinfra, and custom OpenAI-compatible endpoints. Implement provider-specific voice management with automatic voice restoration per provider-model combination, and migrate package manager to pnpm for improved dependency handling.
Key changes:
- Add TTS provider selection (OpenAI, Deepinfra, custom-openai) in settings UI
- Implement provider-specific model and voice lists with dynamic fetching
- Add voice persistence per provider-model combination in savedVoices
- Support Deepinfra models: Kokoro-82M, Orpheus-3B, Sesame-1B with their voice libraries
- Migrate to pnpm with frozen lockfile for reproducible builds
- Update Docker configuration to use pnpm and Deepinfra API defaults
- Add migration logic for existing users to infer provider from stored baseUrl
- Update test helpers and Playwright configuration for Deepinfra API
- Add example docker-compose.yml with Kokoro-FastAPI integration
BREAKING CHANGE: Voice selection is now provider-model specific. Previously saved voices will be migrated to the new savedVoices structure, but users may need to reselect voices if switching providers.
