Based on deep neural network (DNN) hardware processor, enabling local data collection, computation, and decision-making for speech processing without relying on the network. It provides low latency and protects user privacy.
Utilizing deep learning technology to achieve tasks such as echo cancellation, reverberation removal, beamforming, speech enhancement, noise reduction, and sound source localization. This enables clearer speech perception for both humans and machines, facilitating smoother speech interactions.
Noise: 70±5dB, Speech: 70±5dB, Recognition Rate: 90%
Supports angular positioning within ±15° accuracy range in a 180° field.
Enables speech wake-up interrupt functionality with echo suppression ratio exceeding 25dB.
Supports speech separation and directional noise enhancement and suppression.
Eliminates the influence of reverberation from captured sound.
Supports speech enhancement, enabling 360° omnidirectional sound pickup to enhance target speech .
Combines deep learning noise reduction and echo cancellation technologies for high call clarity.
Supports detection of sounds like baby crying, snoring, etc.
Converts speech into corresponding text quickly and accurately
Recognition up to 10 meters away
Fast response time: as low as 0.2 seconds
Fast response time: as low as 0.2 seconds
Languages supported: Chinese, English, Japanese, and some dialects
Identifies individuals by automatically extracting unique vocal characteristics
High recognition accuracy of over 93%
Supports registration of multiple identities simultaneously
Enables real-time identity verification and recognition within a short timeframe
Supports various voiceprint verification modes, including free speech and fixed text
Based on deep neural network technology, it can convert text into clear, natural, and fluent speech in real-time. Provides both offline and online TTS capabilities
30+ voices, 5+ emotions
Natural-sounding voices with diverse styles
Supports TTS in Chinese, English, and Japanese
Supports online TTS on AI platforms
Batch TTS for high efficiency
Offline TTS engine
Meets the need for converting text to speech in offline environments
Recognizes natural language and semantics for quick responses
Supports speech model self-learning. In offline mode, users can customize wake-up words and command words by recording their own speech, achieving personalized and self-service customization
Devices can be controlled through dialects after self-learning.
Suitable for elderly people who don't speak Mandarin.
Customizable wake-up words to differentiate between different products.
Suitable for lighting, air conditioning, and other products.
Customized command words.
Better aligns with user habits.
1
12
23
3Please improve the information. Qiyintalen looks forward to working with you!
