🎙️ VoxSherpa TTS — by CodeBySonu95
VoxSherpa TTS is a professional-grade, 100% offline neural text-to-speech engine for Android. It brings studio-quality voice synthesis directly to your device — no cloud, no internet, no tracking.
Built on Sherpa-ONNX, it runs two neural engines side by side: Kokoro-82M for studio quality and Piper/VITS for speed.
🏆 Featured in the official Sherpa-ONNX README — the core inference library powering this app.
📲 Download
Now available on Google Play:
🧠 Dual Neural Engine
| Engine | Quality | Speed | Best For |
|---|---|---|---|
| 🧠 Kokoro-82M | Studio-grade · rivals ElevenLabs | Slower on budget hardware | Audiobooks, voiceovers, professional content |
| ⚡ Piper / VITS | Natural · clear | Fast on any device | Daily use, System TTS, accessibility |
🚀 Features
- 100% Offline & Private — all processing on-device, no telemetry, no accounts
- System-Wide TTS — set as default Android TTS engine; works in Chrome, WhatsApp, TalkBack, and any app — with pitch & speed control
- PDF to Audio & TXT to Audio — listen to any document hands-free
- MediaStyle Notification — playback controls from notification shade and lock screen
- Emotion Tags —
[whisper],[angry],[happy]inline expression - Smart Punctuation — natural pauses after sentence breaks
- Interactive Audio Seeking — scrub through long audio with mini player
- Chunk-based Playback — audio starts before full generation completes
- Model Management — download in-app or import custom
.onnxmodels from local storage - WAV Export — save generated audio at correct sample rate per model
- Speech Library — history, favorites, timestamps, voice attribution per recording
- 100+ Kokoro Speakers — across Hindi, English, Japanese, Chinese and 50+ languages
🌐 Language Support
Hindi 🇮🇳 · English 🇺🇸 · British English 🇬🇧 · Japanese 🇯🇵 · Chinese 🇨🇳 · French 🇫🇷 · Spanish 🇪🇸 · German 🇩🇪 · Portuguese 🇵🇹 · Italian 🇮🇹 · Korean 🇰🇷 · Russian 🇷🇺 · Polish 🇵🇱 · and 37+ more
📊 Performance
| Device Tier | Kokoro-82M | Piper / VITS |
|---|---|---|
| 🟢 Flagship (SD 8 Gen 3) | ~20–40 sec / min audio | ~5 sec / min audio |
| 🟡 Mid-range (8-core) | ~60–90 sec / min audio | ~10 sec / min audio |
| 🔴 Budget (6-core) | ~2–3 min / min audio | ~20 sec / min audio |
Kokoro-82M prioritizes quality over speed — same architecture powering premium commercial TTS, running fully offline on a mobile CPU.
🛠️ Technical Details
- Language: Java (Android Native)
- Runtime: ONNX Runtime via Sherpa-ONNX
- Engines: Kokoro-82M (FP32) · Piper · VITS
- Audio API: Android AudioTrack (PCM)
- UI: Material 3
- Distribution: Google Play Store
- License: GNU GPL v3.0
- Requirements: Android 11+ · ARM64
📦 Models in This Repository
This space hosts Sherpa-ONNX compatible TTS models for use with VoxSherpa TTS and other Sherpa-ONNX based applications.
Custom .onnx models can be imported directly into VoxSherpa via Models tab → Import Local Model.
🔗 Links
- Google Play: VoxSherpa TTS
- GitHub: CodeBySonu95/VoxSherpa-TTS
- Developer: CodeBySonu95
- Featured in: k2-fsa/sherpa-onnx
VoxSherpa TTS — Because your voice deserves to stay yours.