Abstract: In front of the wave of artificial intelligence, more and more occupations are being replaced

Remember before

There are many lesser-known occupations around us, some you may know, some you may not know, and some you may have heard of but never understood exactly what they do.

Stenographer is such a profession. Many people think that this occupation is simple, as long as can type on the line, in fact, not. To be a qualified stenographer, not only need to have cultural quality, psychological quality, speed quality, but also need to be assessed to obtain the “stenographer professional qualification certificate” to take office. Psychological and ability assessment +300 words/minute speed +97% of the accuracy rate, plus considering the logical relationship between oral speech, so add up the shorthand, you will feel simple?

Today, occupations you don’t know about, such as stenographer, are being replaced. Take the most difficult law industry in the field of sketching as an example, Zhejiang High Court has long replaced the traditional clerk with the help of the judicial voice big data solution. At the scene of the trial, the synchronization record delay of the reporter’s personal test system is not more than 500 milliseconds, and can be automatically corrected from time to time, the accuracy rate is more than 97%. So, what is the product just has such intrepid strength?

Application scenarios

Ali Cloud intelligent voice interaction is based on speech recognition, speech synthesis, natural language understanding and other technologies, endods products with intelligent human-computer interaction experience of “can hear, speak, understand you”. At present, Ali Cloud intelligent voice interaction has been implemented in court shorthand, line detection, intelligent customer service, voice quality inspection, live subtitles and other scenes.

Court shorthand: real-time recording of the entire court hearing, covering more than 300 courts. Example: The people’s Courts of Zhejiang Province.

Line detection: transfer full call to text, find possible phone fraud. Example: Check cloud SaaS products.

Intelligent customer service: traditional customer service to intelligent customer service transformation. Example: Ant Financial 95188 hotline, intelligent customer service robot.

Voice inspection: Checks the service process after the voice is transferred to text. Example: Ali Group customer service, United Life.

Live subtitles: Live subtitles and monitoring. Example: Real-time subtitles of cloud Conference; Austrian point cloud landing cooperation.



Language model self-learning tool

Language model self-learning is an intelligent voice self-learning platform pioneered by Aliyun intelligent voice interaction in the world. It is an exclusive voice model that can help users with zero basic training services.

There are usually some unique words in the business domain. When the default recognition effect is poor, you can consider using generalized words or hot words based on different business scenarios. By adding these words to the word list, you can improve the recognition of these words.

If you have accumulated a wealth of historical data in your domain, you can use this historical data to make custom optimizations to your language model. Through the use of voice from learning tools, can through operational interface to upload the training corpus text, and select the corresponding language in the field of basic model, through the training corpus for model training, can effectively improve the speech recognition rate of this scenario, especially proper nouns and high-frequency words from the text, has a good optimization effect.





Intelligent interactive large screen

One of the major applications of intelligent voice interaction is to be packaged into large intelligent screens that can realize human-computer interaction in various public Spaces. Its most prominent feature is speech recognition in strong noise environment, and it has the ability to understand long sentences without wake up. In March 2018, the world’s first voice ticketing machine was officially launched in Shanghai South Railway Station and Hanzhong Road subway station. Under the real noisy environment of subway, the voice recognition accuracy rate exceeded 96%, and the operation of picking up tickets in 10 seconds with free hands, while it normally takes 30 seconds to pick up tickets manually. At present, the main application scenarios of intelligent interactive large screen are as follows:

  • Big traffic: subway ticket, inquiry, airport, scenic spot, railway station inquiry;
  • New retail: ordering, fitting mirror, fitting mirror, super shopping guide;
  • Government and enterprise hall: government affairs, operators, banks, insurance hall inquiries;
  • Others: hospital triage registration and department navigation, library for books.

Write at the end

Ali Cloud intelligent voice interaction unique voice model training self-learning platform, plus its rich interface types, and experience in telephone, App, political and legal conference field precipitation, to provide developers with great help in intelligent human-computer interaction development.


The original link

This article is the original content of the cloud habitat community, shall not be reproduced without permission.