Ebook Voice Player TTS Engine Deep Dive: Async Synthesis, Smart Caching & Cross-Platform Playback Watchdog

The ebook voice player relies on a sophisticated TTS engine to deliver smooth, robust speech playback. The core components of this engine are outlined in tts_engine.py. The TTSService is responsible for orchestration, managing the state machine, initializing tasks, and handling cleanup. The TTSGenerator handles asynchronous concurrent synthesis, while the TTSPlayer manages playback and incorporates a watchdog for progress assurance.

When a user clicks Play, the TTSService initiates the state machine, clears old tasks, and sets up the audio queue. The TTSGenerator then runs in a dedicated thread with an asyncio event loop, synthesizing each paragraph concurrently with a rate limit protection mechanism in place. A semaphore limits the number of workers, ensuring optimal resource utilization.

The TTSGenerator employs a content-addressed cache strategy, creating a unique cache key based on the text, voice, and rate. If the cache file exists and is valid, it is used for playback. Otherwise, the engine attempts an atomic write using a temporary file, validating the MP3 integrity before replacing the cache file. The TTSService listens for events, enqueues audio paths, and emits paragraph events as the playback progresses.