Source: ATYUN AI platform
July 26, millet new conference as scheduled, millet co-founder Wang Chuan at the scene thanked each of the partners, with particular reference to the sound-Technology (SoundAI). In order to make intelligent speakers are faster and more accurate heard, the experience to do the best, sound-technology provides leading microphone array to wake millet and technical solutions, provides a dual support hardware and algorithms.
The hardware part, millet speakers with sound-Technology annular array 6 wheat, with omnidirectional wake up, the sound source directional, directional pickup, noise suppression, reverb eliminate other functions, to meet all the needs of users in the field of voice interaction.
Wake double name (Dual-wake), wake-up word that is no longer a single one speaker can have two names, using any name can be a wake-up and non-discriminatory operation. If the user called "Little love the students" or "millet TV" can wake the speakers. This design and improved functionality for intelligent speaker linkage between the expansion and implementation of smart appliances is a good start.
Any interruption (Free-cut), the technology allows the speaker at any time immediately respond to the new wake up and terminate the current task, perform new tasks and achieve continuous repetition wake up without having to wait for the machine interval should be on call, and human exactly the same habit. While the technology also optimizes the wake effect task execution devices, such as loud when playing music, the user can successfully wake speakers.
Even after a control (One-shot), a "wake-semantic word + Voice recognition" in an integrated manner to achieve seamless connection between words and wake-up voice control, the user name is the name of the device without waiting for a response device can directly say command.
Founded just over a year, has become widely accepted
Sound-Technology had previously jointly launched cooperation with the size of the water droplets 360 cameras, and other children's stories, but also to establish cooperative relations with foreign ARM, NVIDIA, Xilinx, Rockchip, Cypress and other companies. Not long ago, the company hand in hand Baidu, launched to address the integration of voice interaction far-field DuerOS based program. Today millet help introduce smart speakers, sound-technology can be described as wind and waves, overcoming all obstacles.
In April 2016, when the voice interaction battle only began to take shape, it came from the Chinese Academy of Sciences Institute of Acoustics sound-technology team carefully assess the future prospects of the industry voice interaction, combined with its technological advantages, continue to sound wisdom technology technical solutions, product-depth optimization algorithm adjustment, a series of acoustic perception fusion technology and artificial intelligence technology, develop their talents in their interaction aspects of artificial intelligence AI era the greatest extent.
5 m outer recognized speech command, a microphone is no longer a simple tool radio
In terms of acoustics, sound-technology provided from the chip, module, OS to a series of interactive solutions for cloud services, industry-leading technology, such as echo cancellation, noise suppression, sound source localization, so that the microphone is no longer a simple the radio tool. It does not need to close the device can communicate with them, on the target speaker's voice for the far-field microphone array is enhanced by matching front end processing method and the far-field of speech recognition, so that even still accurately recognizing speech commands from the microphone 5 meters.
Focus on hard science and technology, consumer electronics, interactive voice reconstruction
Acoustic algorithms typically required across the two areas, both to use various algorithms (such as wake-up implementation) Computer Science in, but also repeated grinding test in the laboratory. It is precisely because this area was particularly need some sound like such a hard-technology technology companies.
This switched ATYUN artificial intelligence media platforms, the original link: Sonic Intelligence Technology unveiled Millet new conference
more recommendations
Kasisto: human intelligence to make smarter AI
Cape Analytics to start defensible space size analysis
Deep learning vocabulary (five)
Nature: do not speak, AI brain signals directly translated into speech
