README.md

metadata

title: Ultimate TTS Studio - 900+ Premium Voices
emoji: 🎙️
colorFrom: blue
colorTo: purple
sdk: static
pinned: true
license: apache-2.0

🎙️ Ultimate TTS Studio

900+ Premium Voices from 3 World-Class TTS Engines - All Running in Your Browser!

✨ Features

🎯 3 Premium TTS Engines

1. 🎯 Piper TTS - 904 voices across 50+ languages

   • High-quality multilingual support
   • Multiple quality levels (High/Medium/Low)
   • 3-5x realtime generation speed

2. ✨ Kokoro TTS - 21 expressive voices (Highest Quality)

   • 24kHz studio-quality audio
   • American & British accents
   • Most natural & expressive

3. ⚡ Kitten TTS - 8 voices (Fastest)

   • Only 24MB model size
   • Lightning-fast generation
   • Perfect for quick tasks

🚀 Key Capabilities

• ✅ 900+ Professional Voices - Choose from massive variety
• ✅ 50+ Languages - Speak in any language with Piper
• ✅ Unlimited Text Length - Automatic smart chunking
• ✅ WebGPU Acceleration - Hardware-accelerated when available
• ✅ Zero Server Cost - 100% client-side processing
• ✅ Offline Capable - Works after models cached
• ✅ Privacy First - No data leaves your browser
• ✅ Professional Quality - Up to 24kHz audio output

🎮 How to Use

1. Select Your Engine

For Maximum Variety: Choose Piper TTS

• 904 voices across 50+ languages
• Select quality level (High/Medium/Low)
• Pick language and accent

For Best Quality: Choose Kokoro TTS

• 21 expressive voices
• Studio-quality 24kHz audio
• Perfect for audiobooks & narration

For Speed: Choose Kitten TTS

• 8 fast voices
• Lightweight model (24MB)
• Quick generation

2. Configure Voice

Piper Options:

• Quality: High (22kHz) / Medium (16kHz) / Low (Fast)
• Languages: English (US/GB), Spanish, French, German, Italian, Portuguese, Chinese, Japanese, Korean, + 40 more!
• Top Voices: Lessac, Ryan (US) | Cori, Alan (GB)

Kokoro Options:

• American: Bella, Nicole, Sarah, Sky, Adam, Michael
• British: Emma, Isabella, George, Lewis

Kitten Options:

• 8 voices (Voice 0-7) with different characteristics

3. Enter Text & Generate

1. Type or paste your text (unlimited length)
2. Adjust speed if needed (0.5x - 2.0x)
3. Click "🎤 Generate Speech"
4. Wait for generation (watch progress bar)
5. Play audio or download as WAV

🌐 Supported Languages

Piper TTS - 50+ Languages:

Major Languages:

• 🇺🇸 English (US) - 20+ voices
• 🇬🇧 English (UK) - 15+ voices
• 🇪🇸 Spanish - 30+ voices
• 🇫🇷 French - 25+ voices
• 🇩🇪 German - 20+ voices
• 🇮🇹 Italian - 15+ voices
• 🇵🇹 Portuguese - 10+ voices
• 🇨🇳 Chinese - 10+ voices
• 🇯🇵 Japanese - 5+ voices
• 🇰🇷 Korean - 5+ voices

Plus: Dutch, Russian, Polish, Turkish, Arabic, Hindi, Vietnamese, Thai, and many more!

📊 Engine Comparison

Feature	Piper	Kokoro	Kitten
Voices	904	21	8
Quality	⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Speed	Medium	Medium	Fast
Model Size	~50MB	~80MB	~24MB
Languages	50+	English	English
Sample Rate	16-22kHz	24kHz	16kHz
Best For	Variety	Quality	Speed

🎯 Use Cases

Content Creation

• 🎬 Video voiceovers & narration
• 📚 Audiobook production
• 🎙️ Podcast intros/outros
• 📺 YouTube tutorials

Accessibility

• 👁️ Screen reader alternatives
• 📖 Reading assistance
• 🌍 Language learning
• 📱 Audio content for visually impaired

Development

• 🤖 Voice UI prototyping
• 🎮 Game character voices
• 📞 IVR system testing
• 💬 Chatbot voice responses

🔧 Technical Details

Technology Stack

• Frontend: Pure HTML5 + JavaScript (ES6+)
• TTS Library: onnx-tts-web
• Runtime: ONNX Runtime Web
• Acceleration: WebGPU / WebAssembly
• Audio: Web Audio API

Model Sources

• Piper: rhasspy/piper-voices
• Kokoro: therealtimex/kokoro-tts-web
• Kitten: therealtimex/kitten-tts-web

Browser Requirements

• Minimum: Chrome 90+ / Firefox 88+ / Safari 14+ / Edge 90+
• Recommended: Latest Chrome/Edge with WebGPU enabled
• Features Required: WebAssembly, Web Audio API
• Optional: WebGPU for acceleration

Performance

• Model Loading: 5-15 seconds (first time only, then cached)
• Generation Speed: 2-5 seconds per 200 characters
• Real-time Factor: 3-10x (depending on hardware & engine)
• Memory Usage: ~200-500MB (with models loaded)

💡 Performance Tips

For Best Quality:

1. Use Kokoro TTS for English content
2. Select High Quality in Piper settings
3. Use well-punctuated text
4. Keep sentences moderate length

For Best Speed:

1. Use Kitten TTS for quick tasks
2. Select Low Quality in Piper
3. Enable WebGPU in browser settings
4. Use shorter text inputs

For Most Options:

1. Use Piper TTS for language variety
2. Explore different accents/regions
3. Compare quality levels
4. Try multiple voices for same language

🎬 Quick Start Examples

Example 1: Professional Audiobook

Engine: Kokoro TTS
Voice: Bella (American Female)
Speed: 0.95x
Quality: 24kHz
Text: Your book chapter...

Example 2: Tutorial Narration

Engine: Piper TTS
Voice: Lessac (US, High Quality)
Speed: 1.0x
Quality: 22kHz
Text: Your tutorial script...

Example 3: Quick Announcement

Engine: Kitten TTS
Voice: Voice 4 (Clear)
Speed: 1.1x
Text: Your announcement...

Example 4: Spanish Content

Engine: Piper TTS
Voice: es_ES (Spain Spanish)
Speed: 1.0x
Quality: High
Text: Su texto en español...

🐛 Troubleshooting

Model Loading Issues

Problem: "ERROR initializing" message

Solutions:

• Check internet connection
• Wait for download to complete
• Try different quality level
• Clear browser cache
• Refresh page

No Audio Output

Problem: Player appears but no sound

Solutions:

• Check browser audio permissions
• Verify volume settings
• Try different voice/engine
• Check browser console (F12)
• Test with different browser

Slow Performance

Problem: Generation takes too long

Solutions:

• Switch to Kitten TTS for speed
• Lower quality in Piper settings
• Enable WebGPU (chrome://flags)
• Update browser to latest version
• Close other tabs/applications

WebGPU Not Available

Problem: Shows "WASM" instead of "WebGPU"

Solutions:

• Update browser to latest version
• Enable in chrome://flags → "WebGPU"
• Check GPU driver updates
• WebGPU optional, WASM works fine

🎯 Voice Recommendations

English (US) - Natural:

• Lessac (Piper) - Professional, clear
• Ryan (Piper) - Authoritative, deep
• Bella (Kokoro) - Elegant, sophisticated

English (GB) - British:

• Cori (Piper) - Refined, professional
• Emma (Kokoro) - Elegant, polished
• George (Kokoro) - Commanding, distinguished

Spanish:

• es_ES (Piper) - Spain Spanish, multiple voices
• es_MX (Piper) - Mexican Spanish

French:

• fr_FR (Piper) - France French, multiple voices

German:

• de_DE (Piper) - German, multiple voices

📝 Privacy & Security

✅ 100% Client-Side - All processing in your browser ✅ No Server Upload - Text never leaves your device ✅ No Data Collection - Zero analytics or tracking ✅ No Account Required - Use instantly, no signup ✅ Offline Capable - Works without internet (after cache)

📜 License & Credits

License

This project is released under the Apache 2.0 License.

Credits & Acknowledgments

Libraries & Tools:

• onnx-tts-web by @therealtimex
• Piper TTS by Rhasspy
• ONNX Runtime by Microsoft

Models:

• Piper TTS models by Rhasspy team
• Kokoro TTS by community contributors
• Kitten TTS by community contributors

Inspiration:

• TTS Studio by @clowerweb

🚀 Future Enhancements

Planned features:

• More TTS engines (Coqui, VITS)
• Voice cloning with SpeechT5
• SSML markup support
• Batch processing
• MP3/OGG export
• Voice mixing/blending
• Real-time streaming
• Pronunciation dictionary

🤝 Contributing

Found a bug or have a suggestion? Please open an issue or submit a pull request!

🌟 Star This Space!

If you find this useful, please give it a ⭐ star on HuggingFace!

---

Made with ❤️ for the open-source community

Enjoy creating amazing voice content! 🎙️