Application of the hottest speech recognition tech

  • Detail

The application of speech recognition technology in mail sorting

the application of modern sorting equipment in the postal service has greatly improved the efficiency of mail processing. However, not all mail can be processed on the sorter, and the mail that needs to be processed manually has become the bottleneck for postal enterprises to realize automation. In recent years, postal services have begun to use manual coding technology and advanced computer software systems to deal with mail that cannot be used on the computer, but this still requires a lot of labor, and postal enterprises need to spend a lot of money every year to train employees. In order to save labor costs, postal enterprises try to improve the automation level of manual mail processing

some technologically advanced companies are optimistic about the development prospect of speech recognition technology in manual mail sorting. The main reason why speech recognition technology was neglected in the mail center in the past is that this technology has great limitations, the most obvious of which is that this technology has very high requirements for database capacity and computer performance, and the technical level at that time was difficult to meet this requirement

the database that the early speech recognition system relied on can only recognize the pronunciation of a few words and numbers, and the processing speed is extremely slow, while the required database is extremely large. Such a technical situation makes the speech recognition technology unable to be really applied to production. Although speech recognition technology is still facing old problems, with the development of speech engine and computer performance, speech recognition technology has gradually become an optional technology in mail processing automation

whether in the mail receiving and sending center of a large company or in the post office, using speech recognition technology to process mail requires a powerful database to store and recognize thousands of words and number combinations. For the mail receiving and sending center of large companies, the employee name and email code that need to be recognized by the system are required, and the system is also required to process non-standard mail in a simple way, and can be upgraded frequently. For the post office, the information that the system needs to process includes city and street names, countless delivery addresses, so as to realize the automatic control of the average load, delivery routes and village codes

spell ittm software technology, developed by mailcode and ready to apply for a patent, has revolutionized speech recognition automation equipment by improving the ability of the system database. This technology provides unlimited database capacity and ensures that the sorting speed will not be reduced due to the increase of the database. The system developed by major speech engine companies also supports various major languages in the world. In this way, speech technology has become a worldwide product

taking the English speech recognition system as an example, the system establishes 36 recognizable characters, 26 letters plus 10 numbers from 0 to 9, and also establishes a set of keywords. Spell it uses these characters to recognize thousands of spoken words and countless combinations of words

for mail receiving and sending centers of large companies, using the spell it software technology of mailcode, sorters actually only need to sound a few characters to find the corresponding words in the database. For example, when encountering the mail sent to Joseph Schneider, the operator only needs to sound "J", "s", "C" and "H" to get accurate sorting information. As shown in Table 1, the database will display the following employees: Table 1 and the information corresponding to the letter J S C H

name and email Code: Jennifer Schroeder, software engineering department; Joseph Schneider, technical operation Department; Josh Schriver, technical operation Department, because these three names all meet the pronunciation standard of (J, s, C, H). The operator of the mail center knows that the mail is actually sent to Joseph Schneider, so he can put the mail into Joseph Schneider's mailbox

the post office should distribute the mail according to the delivery route, and the sorter must be familiar with the long list of delivery sections and various international mail delivery information. Spell it technology stores the address, delivery route and other information into the system, which greatly facilitates the sorting work

for example, there is an email that ② forms a deep interface dispersion and mechanical interlock at the interface and is sent to 2036 stonehollow road. Using speech recognition technology, the sorter only needs to send out "2", "0", "s", "t" and "O", as shown in Table 2, and the database will give all possible addresses corresponding to these tones and the corresponding delivery routes. In this example, there are three delivery addresses that meet this voice standard. The sorter knows which is the correct address, so he can distribute the mail to the deliverer of the corresponding delivery section. The database information corresponding to table 2 and 20 sto

spell IT technology can recognize letters, numbers, keywords and countless combinations, which is a complete change for the manual mail sorting of mail receiving and sending centers and post offices of large companies. This progress enables the mail center to use fast, precise and supporting automatic mail processing equipment to improve the efficiency of the whole process of mail processing. Spell it technology can also be applied to many languages in the world

a major problem faced by mail receiving and sending centers and post offices of large companies is how to improve productivity while controlling training funds? What economic benefits can be achieved by reducing the search time for internal mail and mail that cannot be sent to the sorter? Manual mail sorting centers often have to consider labor-related issues: higher and higher labor costs, faster and faster personnel changes, and more and more training costs. Manual sorting in the company's internal receiving and dispatching center and post office mainly depends on the memory of sorting personnel. Recruiting, training and retaining skilled sorters often require high costs. If a skilled worker asks for leave or goes on vacation due to illness or business, or resigns, another person must fill the vacancy. Because the improvement of productivity depends on the proficiency of workers, and many enterprises' frequent personnel mobility has led to the continuous growth of training funds, so it is difficult to achieve the benefit goal

manual sorting mainly depends on employees' memory, while semi-automatic processing such as manual coding reduces the pressure of memory, but it requires a new set of technology. From manual operation to semi-automatic operation, employees often need to completely change their operating skills. When many skilled manual sorters switch to keyboard operation, their productivity does not increase but decreases. This situation occurs when many companies use manual coding stations and short-range or long-distance video coding equipment, because the coding technology completely relies on the keyboard technology of the operator to process mail

compared with coding technology, speech recognition technology can greatly reduce staff training, and can reduce or even eliminate the impact of the transformation process from manual labor to automation on productivity. It no longer requires a lot of work experience to achieve skilled sorting. In fact, the application of speech recognition technology eliminates the need for operators to remember postal information such as mailbox codes. Speech recognition technology is simple and easy to learn, and the staff of the mail center can reach the level of proficiency in only a few minutes. The reduction of training time has contributed to the decline of labor costs. In addition, speech recognition technology improves productivity, which greatly improves the efficiency of the mail center

mailcode has two kinds of speech recognition equipment: accent adaptive equipment and non accent adaptive equipment. Accent adaptive equipment requires operators to carry out accent adaptation training before production. Operators need to spend about half an hour to adapt the equipment to their own voice characteristics, which are saved in the system as files for future use. This device is especially suitable for operators who have local dialects or accents. Non accent adaptive equipment basically does not need voice training, and is suitable for places with more temporary workers and unfixed operators. However, non accent adaptive devices are slightly slower and less accurate than accent adaptive devices

experiments show that speech technology improves productivity. A skilled manual sorter sorts an average of 700 pieces of mail per hour. Such efficiency requires at least three months of training. Using speech recognition technology, an untrained operator can sort 550 pieces per hour in only a week, and 700 pieces per hour in four weeks. In addition, manual sorting can find 150 pieces of mail per hour, while using voice technology, the amount of mail can reach 600 pieces per hour. As operators become more and more skilled in equipment, the accuracy of sorting will be further improved

because most speech recognition systems are developed based on Windows NT with friendly interface, managers can use tool software to track the production situation of operators, mail volume and the production situation of the whole mail center. The latest equipment is also equipped with a database management system, which can ensure the timely update of the database. Whether it is the internal database of the company or the post office database, such a management system can ensure the timely update of address data, so as to ensure that the mail is accurately delivered to the recipient. With these advantages, it is a practical and effective choice to install speech recognition equipment in the mail center to avoid violent vibration

mailcode provides two voice products: voice planning management system and voice sorting system. The voice planning and management system is used to sort mail to the delivery section for the post office, which can sort parcels, letters, flat pieces and mail without postal code or with wrong code once and twice; The voice sorting system is designed for the mail receiving and sending center of the enterprise. It can use the internal multiple databases to generate complete and accurate documents to identify the staff name, department and delivery point at the same time. With the progress of technology, the database of voice planning management system and voice sorting system can be updated at any time, and its speed and accuracy are also getting higher and higher

both the voice planning management system and the voice sorting system use similar hardware. Each equipment is equipped with wireless headphones to eliminate noise, so that the operator can walk freely between the voice sorting equipment. Each sorting table is equipped with a microcomputer equipped with a 33 ~ 43cm high-definition flat LCD. The voice sorting system can also be equipped with a labeler and a printer

with the development of science and technology, speech recognition technology has gradually become a new technology for mail sorting, which can overcome the disadvantages of manual sorting relying solely on the memory of sorters, solve the problem of high personnel costs, and improve the efficiency and efficiency of mail processing. (end)

Copyright © 2011 JIN SHI