My engineering practice topic is based on the large data Q & training Chinese natural language understanding system design, because the method for constructing learn engineering, research study WordNet, frameNet other common semantic network, using a machine learning algorithm study Chinese natural language understanding, build a set of intelligent voice response system-specific knowledge base. Therefore, intelligent voice response system is based on, I chose the analysis 3 models similar software:
1. Microsoft wheatgrass Huna: Search and Bing are based on the depth of neural network technology, products belong to 2C. Microsoft Huna now available in iOS, Android, Window 10 system, she is more like a secretary, you can call, send text messages, e-mail and queries on a Manchester United score. But many less feature in Microsoft Huna domestic phone, voice response is very slow. Microsoft wheatgrass as a dialogue system, settled in micro-channel, microblogging and other applications or Windows 10 system, she can also access a variety of information. Huna with different applications such as wheatgrass can not call the phone, she is more like a virtual character, can accompany users chat (continuous dialogue), playing games (such as artificial intelligence-based image recognition technology).
2. Apple Siri: may we are more familiar with, you can call it is a chat robot applications. Today's Siri addition to the alarm clock, send text messages outside, but also search for photos, play music, offer suggestions.
3. iFlyTek voice: IFLYTEK of "hearing the voice Flying Cloud" including speech synthesis, speech recognition and search, voice dictation technology, the cloud dialect, higher uncommon word speech recognition rate. Hearing the voice flying cloud input method is based on this development. IFLYTEK "primary" to do is to 2B (for corporate users) services will empower their artificial intelligence services to other businesses. In addition, IFLYTEK years have begun to exert 2C (for the consumer) products recently, such as smart speaker, audio Po, voice recorder and other hardware. It can be seen IFLYTEK products are intelligent and more traditional products, rather than a personal assistant or chatbot. In addition, IFLYTEK bias roots "voice recognition" rather than continuous dialogue.
(1) software developers is how to convince you to become their users? Their goal is profitable it? Their goal is to earn cash user What?
Intelligent voice dialogue system to give people great convenience, we can say goodbye to traditional manually opening the phone to enter text query, the search and what we want intelligent robots can talk directly, you can also chat. And robots and voice chat, always bring some laughter, caught my interest.
"Eat" aspect, Siri and Huna can provide you with information about nearby restaurants, it is possible to provide the whole of the secret services to you, such as providing home delivery service, etc., it can be said to be more thoughtful.
"Row" aspect, if you need a cab, then, Siri, Huna will put forward "beyond its capabilities." But it needs to be pointed out that Siri in your proposed destination, the map will jump directly to Apple, and shows the planned route.
"Play" aspect, in recommending the movie, hearing fly voice not only has the latest movie recommendations, as well as various types of film recommendations, incidentally, have some comments, very good. On the other hand, let the voice of the robot sing a song or something, but also quite interesting, and Huna realistic human voice singing songs of the point, quite flavor.
"Chat" aspect, no doubt wheatgrass cheap cheap, more down to earth answers, cause most people get interested, wheatgrass answer is sometimes a bit machine (nei) Chile (han). As wheatgrass sister, Huna performance is not bad, but did not answer wheatgrass more interesting, the other two to slightly generalized.
It is profitable, mainly through licensing fees, service fees, advertising fees profitable way.
How (2) the software is to your hand?
These software usually comes with the phone, usually handset manufacturers and software companies to cooperate, to authorize the installation on the phone.
(3) These software Bug Why? And how to update the new version?
bug before Apple Siri is not subject to the control console can start the program, such as when there is a new arrival time notification can be displayed on the screen there is a new notification, but unlocked to see the details, but Siri bypassing the unlock procedure. Whether the phone is not unlocked, as long as you start Siri, shout it: "Help me read the notice" will be able to hear the details of the notice. This will result in information disclosure, the system unsafe. No hearing fly voice input keyboard sounds, and sometimes not easy, but when the browser search the hearing fly voice can not be ejected. Of course, intelligent voice software is still in its early stages of development, many speech recognition, semantic understanding is still relatively difficult, is not necessarily accurate, this is a common problem in all intelligent voice products. Release an updated version of the information directly through the mobile phone viewing applications, download the latest installation package updates.
(4) When did you start this kind of software is emerging, between how the same type of software is competition? How trends?
Apple Siri appears early, founded in 2007, and other voice products are mostly 14 years began to rise. Now smart voice products moving in multiple directions, endless variety of products, each with its own characteristics and advantages, contending. With the continuous development of artificial intelligence, the future of these products will be more and more, getting better and better.
(5) cited "special" phenomenon you observed when using the software and hardware they any different? These can explain some of the essential characteristics of the software it?
On one occasion, I used Apple's Siri translating "Huawei regressed", the results are displayed in English "Huawei is real bitch", completely wrong, indicating that Apple still has a long way to go, as long as the perfect good itself, will be a good location, We can develop, or to lose a lot of customers. The difference between software and hardware that the software is invisible, hardware is the tangible parts of the machine, the human mind and body is the same, the software controls the operation of hardware. Software and subsystem software and application software, the software is run on a computer program above, but not only the program, can be said to be a collection of programs plus documents.
(6) Your personal use of such software is the first time what time?
The first time use it is probably sophomore, pottery class teacher, when I use the Android phones, Apple mobile phone with the students, often playing with his, iPhone Siri quite interesting. It is genuine.
(7) How do you proficient in this kind of software? It gives you any good, bad?
Often used on the familiar, easy to bring it to life, such as our hands busy to cook, this time need to call, you can use voice systems, voice dialing direct, driving check the map, sing at home, watching movies, etc., direct voice call. The downside is that the semantic identification is incorrect, feedback often wrong information, or do not identify semantics, a waste of time, it is affecting mood.
(8) What you are still using it, or similar software?
Is still in use, I think it is more convenient, and sometimes very needed.
(9) This software is another 10 or 20 years it will continue to exist?
This thing will always exist, and we can not live without voice, speech can not be separated, intelligent voice system can liberate our hands. It is still in the development stage, will be the future development of more intelligent, people can greatly liberate labor, to bring great convenience working life.