iFLYTEK unveils SparkAudio 1.0 speech model and Robot Super Brain Platform 3.0
iFLYTEK used its 2026 Global 1024 Developer Festival to launch SparkAudio 1.0, a speech foundation model combining translation, recognition and semantic understanding, and Robot Super Brain Platform 3.0 for embodied multimodal interaction. The model was trained on 13 million hours of audio and supports 99 languages and 202 dialects.
- SparkAudio 1.0 pairs a 0.65B-parameter audio encoder with a 30B-A3B MoE model
- Trained on 13 million hours of audio, supports 99 languages and 202 dialects
- Trained on Huawei Ascend 910B and Hygon BW1000 chips
- Robot Super Brain Platform 3.0 includes a board, module and embodied AI dev kit
Read next
AI