Skip to content
TheChatArchiveThe Chat Archive
Claudeclaude-3-5-sonnet-20241022· AnthropicHugging Face LMArena Human Preference 140K

What are the SOTA models for TTS in Mandarin (Traditional Chinese)?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-7-sonnet-20250219-thinking-32k’s answer.
  1. User

    What are the SOTA models for TTS in Mandarin (Traditional Chinese)?

  2. Claude Assistant

    Let me share what I know about state-of-the-art Mandarin TTS models as of my last update in April 2024:

    1. FastSpeech 2 with Chinese adaptations
    • Specifically optimized for Mandarin tonal patterns
    • Good prosody control and natural tone transitions
    1. VITS (Conditional Variational Autoencoder with Adversarial Learning)
    • Strong performance on Mandarin
    • End-to-end model with fast inference
    • Handles Traditional Chinese characters well
    1. YourTTS
    • Multilingual model with strong Mandarin support
    • Zero-shot voice cloning capabilities
    • Good handling of tonal variations
    1. Coqui TTS
    • Open source with specific Mandarin models
    • Supports Traditional Chinese text processing
    • Multiple architecture options (Tacotron, FastSpeech, VITS)
    1. Microsoft Azure Neural TTS
    • Commercial solution with high-quality Mandarin voices
    • Good handling of Traditional Chinese characters
    • Natural prosody and pronunciation

    Would you like me to elaborate on any particular aspect of these models? I should note that since this field moves quickly, there may be newer developments after my last update.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.