Dubverse TTS Models
Dubverse offers two powerful Text-to-Speech (TTS) models: neo.one and candy.two. Each model has its unique strengths and is designed for specific use cases. This guide will help you understand the features and applications of both models.
neo.one model
Our fast, efficient TTS model, designed for quick processing and resource optimization across a wide range of languages and use cases.
- Content Localization: Quickly translate and voice-over content for multiple languages.
- News Broadcasting: Generate audio versions of news articles in real-time.
- IVR Systems: Power interactive voice response systems with fast, efficient TTS.
- Bulk Audio Generation: Create large volumes of audio content for various applications.
- E-learning Platforms: Generate audio content for online courses efficiently.
candy.two model
Our flagship model focused on high-quality, natural-sounding speech synthesis, making it ideal for applications requiring premium voice output.
Use Cases
- Virtual Assistants: Create lifelike AI assistants with natural-sounding voices.
- Audiobook Production: Automate the creation of audiobooks with high-quality narration.
- Podcast Creation: Produce professional-sounding podcast episodes from written scripts.
- Gaming: Develop dynamic, responsive NPC dialogues in video games.
- Accessibility Tools: Improve web accessibility with high-quality text-to-speech for visually impaired users.
swift.call model
Our specialized model designed specifically for voicebot applications, featuring real-time voice cloning and streaming capabilities.
Key Features
- Run-time Voice Cloning: Clone voices in real-time for dynamic voice adaptation
- Word-level Timestamps: Get precise timing information for each word in the generated speech
- Streaming Support: Real-time audio streaming for interactive voicebot applications
- Low Latency: Optimized for real-time voice interactions
Use Cases
- Voicebots: Create interactive voice assistants with real-time voice cloning
- Call Centers: Power automated customer service systems with natural-sounding voices
- Interactive Voice Response: Build dynamic IVR systems with streaming support
- Real-time Voice Applications: Develop applications requiring immediate voice feedback
Model Comparison
| Feature | neo.one | candy.two | swift.call |
|---|---|---|---|
| Processing Speed | Ultra-fast | Standard | Real-time |
| Voice Quality | Good quality | Highest fidelity | High quality |
| Resource Usage | Optimized (lower) | Standard | Optimized |
| Language Support | 10+ languages | 8+ languages | Hindi & English |
| Emotion Control | Basic | Advanced | Basic |
| Batch Processing | Optimized | Supported | Streaming |
| Architecture | Non-autoregressive | Transformer-based | Voicebot-optimized |
| Use Case | Efficient, large-scale TTS | High-quality, premium TTS | Real-time voicebot TTS |
| Voice Cloning | No | No | Yes |
| Word Timestamps | No | No | Yes |
Choosing the Right Model
-
Choose neo.one when:
- Processing speed is a top priority
- You’re generating large volumes of audio content
- You need to optimize resource usage
- Batch processing is required
-
Choose candy.two when:
- You need the highest quality voice output
- Advanced emotion and emphasis control is required
- You’re working on applications requiring premium voice quality
- Transformer-based architecture benefits are desired
-
Choose swift.call when:
- You’re building voicebot applications
- Real-time voice cloning is required
- Streaming support is essential
- Word-level timestamps are needed
- You need Hindi or English voice support
For guidance on selecting the best model for your specific use case, please contact our support team at [email protected].