what is new in espeak ng advanced Version 2.10 Consolidates all updates, enhancements, and fixes from 2.10 Beta 1 and Beta 2. 🚀 Highlights & Major New Capabilities • Significantly Faster Speech Startup: Natural voices now initialize and begin speaking noticeably faster across all devices. • Custom Model Imports: Bring your own models—full support for importing Piper and Coqui voice files via sherpa-onnx (Natural voices > Import a voice file), as well as NVDA voice variants. • Dedicated Snapdragon NPU Acceleration: • Near-instant loading on Snapdragon 8 Gen 3, 8 Elite, and newer platforms. • 37 new instant-start Phonon English voices. • Seamless fail-safe: voice playback continues uninterrupted via CPU if the NPU encounters an error. • Download size slashed in half (from 94 MB to 46 MB). • Per-App & Per-Language Voice Controls: • Assign dedicated speed, pitch, volume, and punctuation profiles to individual apps. • Customize individual settings, voices, switching sensitivity, and audio cues per language. • Native Indian Numbering: Indian languages now process numbers natively (e.g., 1,00,000 and ₹500) instead of dropping back to eSpeak. 🎙️ Audio Fidelity & Voice Model Updates • Seamless Transitions: Playback no longer stutters, pauses, or drops back to eSpeak while switching or loading new voices. • Cleaned Audio Stream: Eliminated mid-sentence pops and crackles, normalized volume over long texts, and cut background hiss on Kannada Spoorthi. • Accurate Native Pitch: Voices retain their natural vocal pitch at default settings. • Voice Library Refreshes: • Kannada Vikram & Ananya return with smaller download footprints and NPU compatibility. • SYSPIN & Rasa voices consolidated into smaller, single-package downloads with automatic migration. • Supertonic Voices (Beta): Added 10 voices across 31 languages (optional toggle). 🎛️ Navigation, Automation & Smart Reading • Recent Speech Quick Tile: Replaces Reading History—replays recent spoken text, spells words out, adjusts playback speed, or copies output to the clipboard. • Automation & Presets: Save custom voice profiles switchable via Quick Settings tiles or Tasker automations. • Smart Sleep Timer: Stops reading tasks at scheduled intervals without interrupting system TalkBack announcements. • Expanded Text Parsing (Optional): • Intelligent recognition of dates, fractions, math, temperatures, phone numbers, and Roman numerals. • CamelCase splitting for code identifiers (e.g., getUserName → "get user name"). • Multi-language inline detection for embedded French, German, Italian, Spanish, and Portuguese phrases. • Built-in sound enhancer with live audio preview. 🌐 Language & Pronunciation Fixes • Enhanced vocabulary dictionaries for Hindi, Urdu, Malayalam, and Punjabi. • URLs and email addresses remain unified rather than fragmenting across multi-language switching. • Natural phrasing pauses added after Indian, Arabic, and Chinese punctuation marks. • Restored Swiss German phonetic pronunciations and updated NVDA symbol mapping. ⚙️ Engine Stability & App Experience • Streamlined UI: Essential controls grouped on the primary page; deeper configurations organized under Advanced. • Fixed Settings Search: Quickly jump directly to specific toggles. • Direct In-App Updates: Check and update builds under Backup and about > Check for updates. • Resource Optimization: App package size reduced by ~4 MB with faster cold starts and resolved freeze issues during backup or shared dictionary imports. https://github.com/animeshahilya/espeak-ng/releases/tag/v2.10