On April 26, iFLYTEK's iFLYTEK Xinghuo model launched an updated version of V3.5, which mainly involves new functions such as supporting long text, long graphics and text, and long voice. In addition, new features such as the large model of Spark graphic and text recognition, multi-emotional super-anthropomorphic synthesis, and one-sentence replication are also debuted in this update. These features are mainly aimed at the enterprise application market, which can help users solve real-world scenarios more efficiently and securely.
Why do some of the new features highlight a "long" word, and how does it differ from previous text, graphics, and speech?
In fact, it is not difficult to understand the word "long". As we all know, the development of AI has provided convenience for many office users, and they can quickly generate the content or graphics we need, but when encountering an ultra-long report, paper, etc., if the large model does not support long text, then it may be difficult to complete it manually without eating or drinking for a day, which is the pain point brought by "long".
For professional users such as lawyers, analysts, consultants, marketers, etc., daily work needs to consult and sort out a large number of long texts, so iFLYTEK Xinghuo has absorbed a large number of text materials from various channels this update, and can answer more accurately for professional fields in various industries.
As shown in the figure above, the author uploaded a 11MB capacity of the "Non-ferrous Metal Processing Plant Power Design Code", the document is a total of 130 pages, the whole upload process is very fast, basically no need to wait. At the same time, according to the content in the specification, let iFLYTEK Xinghuo summarize the main content and technical requirements in it, and then wait for more than ten seconds to generate a text summary of the answer.
I believe that any worker will gasp when they encounter a 130-page document, but it is already very difficult to read it, and they still need to find the answer they want in the text, and the difficulty will increase several times. On the other hand, iFLYTEK Xinghuo only needs to upload professional data documents, you can quickly find the content you want to understand, and analyze and summarize it, I believe that it is absolutely heavenly to improve office efficiency.
In addition to the summary of long texts in the professional industry, the new iFLYTEK Xinghuo can also interpret long texts, such as the author uploading a rent lease contract to see if iFLYTEK Xinghuo can find the key information we need.
Contract documents have always been impossible to ignore, after all, it has a certain legal effect, so every time you sign a contract, you will read it carefully, for fear of stepping on the pit. However, using iFLYTEK Xinghuo can screen out the terms that are unfavorable to you, so that after summarizing, you can intuitively see that those terms are not satisfied, and negotiate with the other party as soon as possible, which greatly reduces the risk of stepping on the pit.
In addition, there are some clauses of information that we often do not have the patience to read, such as the "China Pacific Property Insurance Co., Ltd. Motor Vehicle Commercial Insurance Model Clauses", in fact, the purchase of commercial insurance should fully understand its terms, but I believe that not many people can patiently read the dense 16 pages of text.
At this time, iFLYTEK Xinghuo comes in handy, such as the broken glass of the car, the payment of items in the car, the flooded car and other problems, it can be easily found out from the long text, and give the most detailed answers.
In addition to daily work, the new iFLYTEK Xinghuo can also involve reading and entertainment in long text applications, such as the e-books we often download, which can also be analyzed and answered.
Mr. Yu Hua's work "Alive" is deeply loved by the majority of users, and after uploading this 130,000-word PDF document, we conducted a simple test, and some details in the book, including whether Fugui participated in the Liberation War, who he met when he was pulled to be a strong man, etc., it can be quickly answered.
And the understanding of the whole book is also in place, it can interpret the life plight of the people at that time according to the relationship between the characters and the background of the times, and help readers to think deeply.
iFLYTEK Xinghuo's new long picture and text function can be said to be more powerful than the long text function, for example, when we get the instruction manual of a drug, the handwriting is small and many, at this time, we can scan and upload the instruction manual or upload it directly into a clear image to understand what we need.
The picture above is an aspirin drug insert for injection, which can clearly say what adverse reactions will occur after injection, including gastrointestinal, kidney, allergies, etc., and will also list some symptoms corresponding to various reactions, so that we can confirm the current physical condition in time. At the same time, which drugs can not be taken at the same time during the injection can also be clearly listed, which is much more convenient than reading the instructions.
In life, everyone will buy some household appliances and digital products, but they are also too lazy to read the manual, and iFLYTEK Xinghuo can also help us complete the interpretation of the product manual. The above is the electronic manual of the Lenovo E14 notebook uploaded by the author,I asked the most important part of the manual,Then Xunfei Xinghuo listed the most important elements from the manual,For example, battery、Safety risks of power supply,Download some application software,How to connect to the display, etc.,It is really important。
In addition, for some faults, iFLYTEK Xinghuo can also be easily solved. The author asks what to do if the screen is black and the keyboard and mouse cannot be used? It will be very logical to determine the cause of the fault through the elimination method, which is very engineer-like.
We can also upload no more than 100 pictures directly, and iFLYTEK Xinghuo can quickly understand the information of the pictures. The above is some of the Korean travel guides uploaded by the author, a total of 21 mobile phone photos, and then asked what characteristics are necessary to experience if you want to go to South Korea. iFLYTEK Xinghuo immediately collected six categories in the entire picture strategy, including culture, food, leisure and so on.
If you want to travel to South Korea for 5 days, iFLYTEK Xinghuo can also make a detailed strategy, and it is also very logical. Most of them landed in Seoul, so the first day of the trip was in Seoul. Then on the second and third days, it began to expand outward, such as Bukchon Bakyong, Jeju Island, and so on. Then there is the fourth day of shopping and entertainment experience, and the last day of leisure return, it can be seen that the summary and logical ability of iFLYTEK Xinghuo is very powerful.
Not only that, iFLYTEK Xinghuo's long graphics and texts can also be applied to education, newspapers, experience reports, app screenshots and other scenarios, for example, students and friends can take pictures of daily teachers' board books, go home and upload them uniformly, and then they can summarize the key content of the day for review.
For journalists, listening to interview recordings on a daily basis is commonplace, and sometimes hours of recordings need to be listened to back and forth many times, resulting in a lot of repetitive work. After using the long voice function of iFLYTEK Xinghuo, you only need to upload the audio file to get the key content you need at any time.
The above is an audio file uploaded by the author of an interview with Zhang Xingang, a professor, master's supervisor and doctoral supervisor of Shandong University, which lasts about 1 hour. The subject matter mainly revolves around the spiritual temperament of modern people and modern Chinese that he observes, what kind of state and what mechanism is generated.
It is not difficult to see that the work efficiency of iFLYTEK Xinghuo is very high, the whole upload process is very smooth, and the questions can be answered quickly after the questions are raised, which shows that the audio to text work should have been carried out when the audio file was uploaded. And it is also very good for induction, for advice to young people, it summarizes 7 main points according to the entire audio file, which is still very in place.
During the whole process of our test, all uploaded documents will be recorded in the personal space, which is not only convenient for us to find, but also to enrich the personal data resource library, so that we can conduct more comprehensive Q&A.
The updated version of the iFLYTEK Spark large model V3.5 also launched two major functions: multi-emotional super-anthropomorphic synthesis and one-sentence voice reproduction. One of the sentences is that you can record your own or other people's voices by reading a dialogue, so that when you communicate with iFLYTEK Xinghuo artificial intelligence, it is like chatting with yourself or friends, and the intimacy is instantly improved a lot.
And after the multi-emotional super-anthropomorphic synthesis added by iFLYTEK Xinghuo, it is more like a conversation with a real person, we can feel its happiness, sorry or coquettish, and the tone is completely different from the usual AI, making communication more pleasant.
summary
Under the premise of the blowout development of AI artificial intelligence, the launch of long text, long pictures, and long voice has undoubtedly enhanced the core competitiveness of iFLYTEK Xinghuo large model V3.5, and also broadened its user field. For some enterprise users who often have to deal with tens of thousands of words or even hundreds of thousands of words of documents, long text is undoubtedly the straw that saves lives, even if the industry is relatively biased, it can still find the answer you want from the uploaded information, which is the biggest core value for users.
It can also be seen from the tests of the above cases that the updated iFLYTEK Xinghuo large model V3.5 is very strong in the recognition and understanding ability of long text, long pictures, and long voices, because the text is too long and the content is too scattered, if you want not to miss key information, you need to have good technical algorithm support, there is no doubt that the iFLYTEK Xinghuo large model V3.5 is very qualified. At present, all the functions mentioned above have been launched, and users in need may wish to try it for themselves to see if it can bring convenience to your work in your field.