On January 7, 2022, Microsoft officially announced that with the support of TTS (Text-to-Speech), a deep neural network based on Microsoft's intelligent cloud Azure, Xiaopeng Motors, a leading intelligent electric vehicle company in China, has successfully completed the upgrade of its vehicle-level voice assistant, further improving the technical level of intelligent car voice assistants.
At present, Chinese users who purchase the P7 model of Xiaopeng Motors can upgrade the new intelligent voice assistant "Xiao P" that can compete with the beautiful voice through OTA (Over-The-Air), and Xiaopeng Motors also plans to introduce this technology upgrade to several other models through OTA.
Thanks to Microsoft's research in the fields of speech, natural language and machine translation in the past few years, voice assistant technology has greatly improved in fluency, quality, fidelity and naturalness.
After integration with Microsoft Azure AI technology and other products, these technological innovations have successfully empowered companies like Xiaopeng Motors to bring a richer and more engaging user experience to their consumers.
In the months of cooperation, Microsoft and Xiaopeng Motors have joined hands to overcome three technical challenges faced by the application of speech synthesis technology:
First of all, in order to solve the problem of network jitter in the automotive scenario and ensure the continuous operation of the high-quality voice function, Xiaopeng Automobile has built a multi-level cache architecture, which can preset and cache high-quality voice files in advance, which minimizes the dependence of this function on the network;
Secondly, in order to provide a real sound experience comparable to that of humans without occupying too many resources, Xiaopeng Automobile can compress voice files to a sample rate of 24KHz and a quantization level of 16 bits with the help of the caching and compression functions of Microsoft Intelligent Cloud Azure, which greatly reduces the resource pressure of the data network and the computing power of the whole vehicle;
Finally, the two sides have made a number of improvements in reducing synthetic speech ambiguity and optimizing accuracy for polysyllabic words.
With the efforts of both parties, the new in-vehicle speech synthesis function has reached a new level in voice fidelity, functionality and scene optimization, and Xiaopeng Automobile can deploy voice assistants in more use scenarios, making it an indispensable part of the intuitive driving experience.
Hao Chao, senior expert of AI products at Xiaopeng Automobile, said: "From determining the intention of cooperation to the launch of the product, we spent several months with Microsoft to complete a cutting-edge exploration of automotive voice interaction technology, raising the natural voice level of in-vehicle voice to a new level. With the deepening of the understanding of urban mobility and the exploration of more use scenarios, these technological achievements will be more widely used to achieve a high level of human-air co-driving experience. ”
Sanjay Ravi, general manager of Microsoft's automotive, mobility and transportation industries, said: "As research and technology advance, Azure cognitive services such as vision and voice will play a key role in defining unique in-vehicle experiences. Intelligent voice is becoming the main in-vehicle interaction tool, and Microsoft's preset deep neural voice and personalized deep neural voice customization services will help automakers strengthen their brands and create a differentiated and authentic user experience that is closer to the natural voice. ”
In addition to Xiaopeng Motors, Microsoft has also carried out in-depth cooperation with a number of automakers and partners in the field of smart cars, focusing on promoting the intelligent application of the automotive industry.
Different manufacturers have different intelligence needs, from the perspective of human-computer interaction to driving information analysis, judgment, and decision-making, different brands and vehicles need to load intelligent applications with different needs.
Based on the underlying platform of powerful speech semantics and data architecture, Microsoft empowers many intelligent car manufacturers through strong technical capabilities and the underlying platform, develops a central control display voice system with diverse forms of information and data, and cooperates with multi-dimensional hardware structure to make the user experience a more intelligent cockpit interaction experience.