iFlytek has officially launched its new SparkAudio 1.0 foundation model during the Global 1024 Developer Festival. The model integrates advanced capabilities including machine translation, speech recognition, and semantic understanding to enhance core audio processing tasks.
SparkAudio 1.0 features real-time speaker diarization, environmental context analysis, and emotion recognition. The company demonstrated the model's ability to identify audio content, background noise, and speaker gender through movie clip analysis.
The new model is now available across several iFlytek products, including the iFlytek Input Method 16.0, iFlytek Hear, iFlytek Office Tablet, and iFlytek Translator. It supports seamless switching between Chinese and English languages.
Additionally, iFlytek introduced the AI Recording Ring hardware device. This gadget features a 4+1 microphone array and Wi-Fi 6 connectivity. The device claims a recognition accuracy rate exceeding 96 percent and can transcribe two hours of audio in two minutes.
