Developed in collaboration with central government agencies and professional research institutions. Uses millions of national-level Mandarin, English, Taiwanese, and Hakka language data to build the most accurate local language model. Employs LLM and word extraction technology to automatically learn industry terms and common phrases, ensuring the model stays current.
|
AI |
Competitors | |
| Language Recognition |
Supports mixed Mandarin, English, Taiwanese, and Hakka recognition. | Primarily English-focused, with lower Chinese recognition capability. |
| Model Capability |
Custom model training understands enterprise-specific terms. | General models cannot handle specialized terms. |
| Recognition Accuracy |
Up to 97% ASR accuracy after tuning. | Around 85%. |
| Template Flexibility |
Custom summary formats with 15 scenario templates. | Fixed formats only. |