On December 6th 2018, at the China Mobile Global Partner Conference, CMCC Smart Speaker S1 and mini S1, as the very first smart speaker supporting phone communication in China, powered by SoundAI Azero, an intelligent interaction system developed by SoundAI, were officially launched. CMCC Smart Speakers were built in a full set of intelligent speech interaction experience and services of SoundAI Azero, keyword spotting, ASR, NLP, TTS and personalized content and skills development.
Huawei AI Speaker, unveiled on October 26th 2018, is powered by SoundAI far-field intelligent speech interaction engine, an integrated solution of hardware and software based on 6-microphone ring-shaped array, with functions of far-field Real-time Call and Voiceprint Recognition. Huawei AI Speaker, is the very first AI speaker in China that supports VoIP, driven by. SoundAI Azero , a whole-chain far-field intelligent speech interaction system, integrating acoustic wave network configuration, beamforming, sound source location, noise suppression, reverberation cancellation, echo cancellation, speech wake-up, voice activity detection, speech recognition, voiceprint recognition, semantic comprehension, speech synthesis, duplex communication, nature language processing and other algorithms, which can ensure 5-meter far-field accurate speech wake-up and recognition under noisy environment and enable unobstructed human-machine interaction experience in real world.
On November 2018, GOME smart speaker, GOMEPOD, powered by SoundAI intelligent interaction system, SoundAI Azero, was officially launched. Based on SoundAI 6-microphone annular array technology, powered by intelligent interaction system SoundAI Azero speech awakening algorithms, including sound source location, beamforming, noise suppression, Open AEC, AKS, and echo cancellation, GOMEPOD can realize 360-degree omni-direction voice pick-up and barrier-free far-field speech interaction in real home environment. Besides, SoundAI Azero far-field speech recognition speech, adopts Bayesian learning framework to extract features of speech data, establish acoustic and language model based on neural network, generate dynamic recognition optimal solutions, and improve word and sentence recognition accuracy by expanding data pool and continuous optimizing models. SoundAI has also deployed scenario task recognition data system to consolidate scenario feature training and learning combined with user habits, enabling GOMEPOD with more accurate speech recognition and reasonable interpretation.
Xiaodu Smart Speaker is a hands-free AI speaker unveiled by Baidu on June 11th 2018. Xiaodu Smart Speaker is powered by the unique SoundAI Azero technology, which includes the world’s first triple-microphone array, far-field voice awakening, AI Voice Activity Detection (VAD), and far-field Voice recognition technology. The product also supports the unique AKS, vertical anti-noise recognition (VAN), and OpenAEC technology of SoundAI. SoundAI Azero keeps the latency between the consumer’s question and the device’s answer within 1.4 seconds, and 400 to 500 milliseconds as for the machine’s response. It supports multi-turn dialog under geek mode as well as vertical voice recognition optimization for children. SoundAI helps Baidu to find balance between a better user experience and an optimized product cost for its product and ensure Xiaodu smart speaker adapts to diverse complicated scenarios while still maintain high performance in far-field speech interaction under different complex scenarios.
MI AI speaker was released on July 26th, 2017. SoundAI is the sole designated supplier of far-field voice interaction solution for MI AI speaker which includes 6-microphone ring array technology and far-field awakening technology with dual-wake, Free-cut, One-shot and other unique customized functions.
MI AI Mini speaker was unveiled on April 3rd, 2018. SoundAI provided 4-microphone far-field voice interaction technology for it, solving the obstacles of small size (microphone array is too close with the loud speaker) and large distortion of cheap speakers. It has a good balance between the product costs and the user experience. By using the latest algorithms, we achieved that the MI AI speaker could still maintain a very high awakening rate and recognition rate despite of various complicated scenarios and excellent adaptability.
SoundAI provides microphone array of mass production, far-field interactive technology, including beam-forming,noise suppression,array gain, echo cancellation，speech arousal,speech recognition and other features, which effectively shield noise interference and achieve long-distance human-machine dialogue.
Lenovo MINI AI speaker officially unveiled on June 13, 2018. It brings you full-featured home experience, all the equipment in the field can "understand your words", you can freely give orders to Lenovo MINI AI speaker equipment; open the devices through AI speakers to connect and work together to give you the most comfortable Experience! SoundAI has exclusively provided 3 microphone far-field voice interaction technology for it to solve the difficulties faced by small smart speaker—small size (microphone array is very close to the loud speaker) and large distortion of low cost speakers. It has a good balance between the product cost and the user experience. Through the use of the latest algorithms, SoundAI has guaranteed that the Lenovo MINI AI speaker can still have a very high awake rate and recognition rate in a variety of complex scenarios because its outstanding far-field pickup capability gives it a good scene adaptability.
VIOMI V Speaker was released on March 9th, 2018. Its far-field interaction capability was provided by SoundAI. The microphone array can support intelligent speaker with screen with the latest voice interaction technology such as Free-cut, One-shot, SSP, and SSA.
The Alibaba Magic Box was released on May 27th, 2018 at the press conference of Shenzhen Satellite TV. For this product, SoundAI implanted technologies like OpenAEC, VAN, and AKS, which are specially designed for intelligent STB, aiming to solve the echo cancellation issues (self-noise suppression) in complicated scenarios such as weak reference signals or even no reference signal. These technologies adapts to various sound effects as speaker, stereo, and surround of different brands or models of TV, in that way ensured more sensitive wake-up and accurate recognition.
Sengled Smart Speaker Lamp was released on January 8th in Las Vegas, USA. SoundAI provides microphone array from volume production, far-field interactive technology and Baidu DuerOS AI system for this product.
On November 12th, Tiny Mu Intelligent Toilet Seat, a product of Xiaomi Eco-chain, powered by SoundAI far-field intelligent voice interaction system SoundAI Azero , was officially launched. SoundAI has tailor-made an integrated solution combining online and offline far-field awakening and recognition speech interaction technologies, through algorithm optimization, even in the network offline state, which can realize local offline voice command control without awakening by using shortcut words. It has small computation volume and memory usage and can respond to complex applications with improved accuracy, which guarantees smooth operation of local memory space and eliminating the impact of network signals influences on natural human-computer interaction.
On October 28, and December 12, 2018, King of Glory Intelligent Robot in shape of game character, Lv Bu, and a female character, Sun Shangxiang, went on the market respectively, powered by penetrated far-field intelligent speech interaction solution innovated by SoundAI. It includes Tailor-made Inverse Microphone Array technology,further enhancing sound signal processing capabilities and resisting acoustic diffraction and reflection caused by the humanoid robot, Penetrated DOA technology, suppressing self-noise and other external noises and improving the accuracy of the sound source location with 10 degrees error of 360-degree omnidirectional location, even under strong noise interference, reverberation and reflection Dynamic Sound Field Vibrated AEC technology, effectively improving wake-up rate in the state of music or Text to Speech, as well as 5-meter far-field speech wake-up and recognition rate under a noisy environment.
Qihoo 360 released its Children’s Robot on July 20th, 2016. SoundAI provided a low-power single microphone recognition solution with noise reduction and duplex communication for it. The recognition of this robot is over 90% within one meter which is a great advancement of children’s voice recognition.
SoundAI provides 360 Smart Camera consumer-grade acoustic solutions of intelligent security, including duplex communication, speech recognition, crying recognition, abnormal sound detection, and other functions. The camera was installed with duplex communication and speech monitoring devices which can help label the monitoring screen, while on the other hand, the employment of deep neural networks makes it possible to achieve far-field awakening and active sound detection all day, once there is anything wrong or the baby cries, messages will be sent to parents.
The 360 Smart Camera launched by Qihu 360 adopted the speech processing chip by SoundAI with functions like double-talk, articulatory suppression resistance, and double mode of speech recognition and call.
The 360 Smart Story Machine was launched on May 18th, 2016 by Qihoo 360. For this product, SoundAI provided single microphone noise reduction recognition and duplex communication algorithm. With the software SDK mode granted through Internet, toy manufacturer have no need to add extra hardware while the toy experience is better.