Interview with Cui Zouzhao, the screenwriter of "The Fox Wife in the Mountain Village": The creation is awe-inspiring, and the pen is touching people’s hearts (2372)

  Wen Yi Yi Jun

During the years of the Republic of China, there was a chaotic world. On a rainy night with a dark wind and a high wind for a month, in an isolated ruined temple, criminals, officers, blind storytellers, women and children, actors and other people with different intentions talked about the legend of "the fox wife in the mountain village". Through the different versions of "Rashomon" style, the mystery of a period of 15 years ago was uncovered layer by layer, and a terrible conspiracy was quietly staged …

Recently, based on folk legends, online movies have incorporated many elements such as thriller, suspense and love.Fox wife in mountain villageBecome a new explosion. This part consists ofStrange trees have fish.Pingzhi movie and televisionBeixia picturesThe works produced not only have a good reputation, but also have a high score in Douban.6.5, set a record for the highest score of the online movie thriller Douban, and the cumulative box office will soon break through the 20 million mark. The film’sScreenwriter and script plannerCui Zouzhao, once again attracted the attention of the industry.

 

The pen is to the heart, and the core of the story is always people.

As a screenwriter of a series of folk legends such as The Legend of the Hunter in Xing ‘an Mountains, Cui Zouzhao said frankly that the creative process of The Fox Wife in the Mountain Village is a brand-new challenge. Although the word "fox wife in mountain village" is full of curiosity and dramatic tension, it is easy to fall into the stereotype of rural tales of mystery. How to solve the problem is often the first and most difficult challenge for a work.

"Every time there is a new project, we must strive to innovate and break through. This process is both painful and exciting. "Cui Zouzhao didn’t want to write easily, but kept thinking and refining the core of the story. He read a lot of folk information, studied fox fairy stories around the country, and gradually sorted out a set of folk story logic related to "fox".

Screenwriter and screenplay planner Cui Zouzhao

"In fact, whether it’s Strange Tales from a Lonely Studio or folk stories from all over the world, the legend about’ fox’ is often based on people’s desire and expression for morality and human relations. Curious and strange stories are just appearances. In the final analysis, everything still lies in people themselves." In terms of narrative techniques, he firmly believes that we should respect the classic drama logic, not deliberately be unconventional and violate the narrative logic, and at the same time, we should constantly innovate, combine local folk customs and audience expectations, and create a paradigm that conforms to contemporary aesthetics and has profound expression. "In story creation, although we can look back on some classic structures, we can’t copy them mechanically. Creating new expressions from classic content may be a part that every creator is challenging himself. "Cui Zouzhao said.

Based on these thoughts and studies, Cui Zouzhao began to try to outline the embryonic form of such a story: starting with the mysterious spiritual characteristics of the fox and putting these characteristics on the human body, what kind of identity would such a person be? Assuming that "the fox wife in the mountain village" is a rumor, what kind of secret is hidden behind it? If it is in such an extreme environment as "Blizzard Mountain Villa Mode", what kind of suspense situation will it cause if someone tells the story of the fox demon like a storyteller in the past …

The train of thought opened like a spring, and Cui Zouzhao began to create vivid characters, taking their inner motives and ups and downs as the story engine of The Fox Wife in the Mountain Village. With the construction of word by word, a subtle narrative maze has sprung up. On the surface, it is the story of "the fox wife in the mountain village" told by several people, but the real protagonist of the story is Li Shouting, a criminal who is full of crimes and treacherous, and Xie Jin, a squire who has never really appeared in the story.

The multiple narratives of virtual reality and virtual reality are very enjoyable, and the story is reversed layer by layer, which is unexpected and reasonable. Talking about how to integrate such a high-density group drama, multiple clues and diverse styles into the limited creative time, Cui Zouzhao said frankly that there is no secret, and it is nothing more than "pen power to people’s hearts".

"The core of the story is always people, exploring human nature, and there are the brightest, most frightening and most complicated things in it. For example, at the beginning, my setting for each character actually echoed the setting of the seven deadly sins. When I was creating, I also consciously put the camps on both sides in different light, one side was cold and the other side was warm, which actually showed the change of people’s state. The conflict is in the minds of the characters as well as the audience. The barrier inside and outside the story is broken by creation, and the attraction of the story naturally arises. But in any case, we still convey the light of human nature. "Cui Zouzhao said.

Create with awe, and the audience has a steelyard in their hearts.

After the launch of The Fox Wife in the Mountain Village, Cui Zouzhao gained a sense of satisfaction and value as a practitioner and creator after watching the heated discussions and various wonderful analyses and conjectures of many audiences on platforms such as Douban. "For the creators, it is undoubtedly a recognition and a kind of luck to attract the audience to participate inside and outside the story. "It has been Cui Zouzhao’s" careful thought "for many years to attract the audience into the thinking and re-creation of the story through his works.

As many people know, more than ten years ago, Cui Zouzhao was already a popular online novel writer. Cui Zouzhao said that his years of writing experience made him understand more and more that he must understand the thoughts of readers and audiences, and respect them and let them integrate into the story, so as to achieve creative resonance.

"The more in-depth creation this line, in fact, the more I panic, the more awe. Our readers and viewers are excellent and know what they want. We must write step by step and make better and better works. " In Cui Zouzhao’s view, fast-paced, high-density and overwhelmed content must return to the story itself. "A good story is always the first. If the story can’t impress people, there is almost no value in redevelopment."

In order to ensure that his stories are getting better and better, Cui Zouzhao can be described as "trying his best". For example, as long as there is an opportunity, Cui Zouzhao must follow the set all the time, which is not only to revise the script in time to improve, but also to make concerted efforts with the creative team to seize every opportunity to make the shooting more perfect. "It’s like the last egg in the film. In fact, shortly before the boot, I chatted with director Liu Xuandi and actor Guo Donghai. How to make the story look better and make the film more colorful, we have to work harder whenever possible."

Cui Zouzhao knew very well that folk culture is a very large system, and the north and the south are different from each other, so it is absolutely impossible to make it up. To be a good storyteller, we must implement folk customs. Therefore, he personally went deep into the countryside, forests and even scenic spots, and personally inspected and accumulated relevant materials. "On one occasion, we went deep into the tomb of a scenic spot and felt the true original appearance of the tomb. The feeling of terror and pressure was particularly shocking. I knew what it felt like to write such a theme." Cui Zouzhao firmly believes that texture comes from real life, and it can only be reached if it is understood in place.

For Cui Zouzhao, it is a "walking on thin ice" to turn his own words into movies. "I feel really lucky to be able to take what I like as a career, so I cherish and fear it all the more. I know that writing a novel is a matter for one person, and making a movie is a battle for a group of people. This process is very happy and serious. You should be responsible for your team and the audience who come to see your work. The audience has a steelyard in mind, and they can clearly distinguish between good and bad, and they can’t be careless in all aspects. “

In order to make stories look better, Cui Zouzhao’s stories often have multiple meanings and multiple interpretations. "There may be more unexpected hidden endings. For example, there are other mysteries about the identity of blind storytellers … anything is possible." Cui Zouzhao extended a meaningful invitation to the audience, expecting everyone to participate in more interesting exploration of the story. "The ideological core of the film must not be imposed on the audience, but guide everyone to go deep into it with plots and emotions. This subtle communication is effective."

Take a new road, do multiplication and addition well, and then talk about subtraction.

The fox wife in the mountain villagedirectorLiu xuandi, and Cui Zouzhao are not only staunch comrades-in-arms who have cooperated for many years, but also like-minded creative friends. When talking about their views on the online film industry, the two people have the same view: "Do multiplication and addition well first, and then consider subtraction."

"It must be admitted that many viewers have inherent prejudice against online movies and some stories that are bloody and shoddy. Our team is very lucky. Companies like Wonder Tree and Fish, Pingzhi Film and Television give us full trust and support, so that we can create freely on the track in the same field to the greatest extent. Then we should do story multiplication and production addition in creation. Of course, subject to many factors such as the viewing experience of small screen and big screen, production expenses and admission methods, the sub-items have always existed, but we must do what we can to the extreme. " Cui Zouzhao said.

When it comes to the breakthrough of this film, Cui Zouzhao thinks that he and his team are trying to find a new way about domestic thriller movies. "At the moment, there are already many different roads. At this time, we have to think about whether to change the road and have a new scenery. The number of modeled roads is always only a temporary scenery. Our audience is not simple, so we must always keep fresh and interesting when the audience has figured out the creator’s routine, without losing profound expression."

In order to achieve the expected effect, Cui Zouzhao not only skillfully used the "Chekhov’s Gun" in the script, but also set the gun hit the floor by villagers and adopted sons from story suggestion to reality. At the same time, we will strengthen the imaginative scene design that can be realized in actual shooting, and fully consider the production cycle and the difficulty of special effects, and do our best to make the story play its greatest role. For example, in the scene of Fox Fairy’s Cave House in the film, Cui Zouzhao and the director designed several schemes together, and finally chose a giant fox head wrapped in vines to show the mysterious visual impact, which impressed the audience deeply. For the expression of thriller scenes, he also tries to use the sense of scene to create psychological feelings. "My definition of terror stems from people’s awe of the unknown, so I will design something that resonates with everyone in life."

When talking about thinking about the future of online film creation, Cui Zouzhao said that we must always push ourselves to a dead end and not stay in the same place for a moment. "After years of running-in, our team has been thinking about how to present the story to the best effect in a limited budget, and I have become more and more efficient and pragmatic in the process of understanding the whole process. This is our advantage, and we should continue to walk out of a newer and better way on the premise of maintaining it. “

It is not difficult to find that the current online movies have gradually changed from winning by novelty-seeking commercial selling points to the stage of relying on high-quality content for long-term development. Cui Zouzhao said that he hoped that when he was getting old, he would not feel remorse when he looked at today’s works. "Creators have to give themselves an explanation. As long as they can write better, they can’t lie flat."

For the future creative planning, Cui Zouzhao said frankly that he was also preparing the content related to the network short play, and called on more outstanding creators and teams to walk side by side. "We also expect more people to see us. Whether it is online movies or online short plays, the support of each platform on each track is getting bigger and bigger. Those who do a good story with their heart will eventually have a better tomorrow."

Editor | Weiss

 

A number of policies have exerted positive factors to accelerate the accumulation of economic operation and continue to move from "stable" to "good"

CCTV News:On September 24, financial institutions announced a number of heavy financial policies: lowering the deposit reserve ratio and unifying the minimum down payment ratio of mortgage loans to 15%, etc. Financial "living water" further supported steady economic growth. High-quality development in the field of people’s livelihood has also achieved inspiring results, bringing tangible happiness and gain to the people; In addition, "News Network" also reported the good news of autumn grain harvest, aerospace science and technology, and drug supervision. The achievements in various fields have boosted confidence, all parties have maintained a good momentum of development, and the "stable" situation of China’s economic operation has not changed.

A number of policies have further supported steady economic growth.

The Press Office of the State Council held a press conference on September 24th, and the principals of the People’s Bank of China, the General Administration of Financial Supervision and the China Securities Regulatory Commission introduced the financial support for high-quality economic development.

Pan Gongsheng, governor of the People’s Bank of China, said that the deposit reserve ratio will be lowered by 0.5 percentage points in the near future, providing long-term liquidity of about 1 trillion yuan; Reduce the policy interest rate, and reduce the 7-day reverse repurchase operating rate by 0.2 percentage points; At the same time, it will drive the medium-term lending convenience interest rate, loan market quotation interest rate and deposit interest rate to go down simultaneously. Guide commercial banks to reduce the interest rate of existing mortgage loans to near the interest rate of newly issued loans, benefiting 50 million households with a population of 150 million, and reducing interest expenses by about 150 billion yuan every year. Unify the minimum down payment ratio of the first and second home loans at the national level to 15%. For the first time, two new monetary policy tools were created to support the stable development of the stock market, namely, 500 billion yuan for the convenience of exchange of securities, funds and insurance companies, and 300 billion yuan for stock repurchase, overweight and refinancing, and the scale could be expanded as appropriate.

Li Yunze, director of the General Administration of Financial Supervision, said that efforts will be made to open up blocking points, improve economic and financial adaptability, and increase financial services for key areas and weak links. Strengthen the financing guarantee of "double" and "two new". At the same time, improve the ability of large commercial banks to serve the real economy.

Wu Qing, chairman of the China Securities Regulatory Commission, said that it will actively support listed companies to conduct mergers and acquisitions around strategic emerging industries and future industries, and vigorously support listed companies to transform and upgrade in the direction of new quality productivity. At the same time, the CSRC and other departments have formulated guiding opinions to promote the entry of medium and long-term funds into the market.

Insist on safeguarding and improving people’s livelihood in development

The State Council Press Office held a series of press conferences on "Promoting High-quality Development" on September 24th, and the relevant person in charge of the Ministry of Human Resources and Social Security introduced the achievements of high-quality development in employment, social security and other fields.

According to the relevant person in charge of the Ministry of Human Resources and Social Security, since the 18th National Congress of the Communist Party of China, there have been 13 million new jobs in cities and towns in China, and the number of migrant workers out of poverty has remained stable at more than 30 million, maintaining the overall stability of the employment situation. Take promoting high-quality full employment as the priority goal of economic and social development, actively promote the synergy between employment policy and macro policies such as finance and currency, and improve the employment driving force of development. We will improve the lifelong vocational skills training system, deepen the Skills China Action, and launch the Skills Strengthening Enterprise Action to enhance the employability of workers.

In terms of social security, since the 18th National Congress of the Communist Party of China, China’s social security system has entered the "fast lane", made historic achievements and made historic changes, and successfully built the largest social security network in the world. The number of people participating in basic pension, unemployment and work-related injury insurance reached 1.074 billion, 245 million and 301 million respectively. The scale of fund revenue and expenditure and balance has increased steadily, and a five-level management service network has basically taken shape in urban and rural areas, with social security cards covering 98% of the population.

Autumn grain has ushered in a large-scale harvest.

Autumn grain is the bulk of the annual grain production, accounting for three quarters of the total grain output. At present, China’s autumn grain has ushered in a large-scale harvest.

These days, the rice in Heilongjiang has been harvested. In Fujin city, with the roar of machinery, the harvester shuttles back and forth in the rice fields, harvesting, threshing and crushing rice stalks in one go. Autumn crops such as corn and soybeans in Northeast China are also about to usher in a large-scale harvest. Sichuan invested about 65,000 sets of autumn grain harvesting machinery this year, and the autumn grain harvest has exceeded 80%. At present, the local corn harvest is basically completed, and the rice harvest is coming to an end.

The central plains granary ushered in a bumper harvest. At present, the autumn harvest in Henan has passed 40%. In Luohe, more than 1.44 million mu of corn harvest is busy, and the local unmanned harvester is promoted to achieve high-precision operation throughout the whole process.

The application of a series of new technologies and varieties has also laid a solid foundation for the harvest of autumn grain. Potatoes in Dingxi, Gansu Province have been harvested in a large area. This year, the local authorities vigorously promoted the integration of water and fertilizer and efficient water-saving irrigation technology. Potato planting saved about 30% of fertilizer and 40% of water, and the yield per mu doubled. Guizhou vigorously promoted small and medium-sized agricultural machinery suitable for mountain use this year. In Shiqian county, a small harvester was used for the first time in the steep mountain slope that could only be harvested manually in the past, and the harvesting efficiency was improved by more than five times. More than 30 new varieties of saline-alkali tolerant sorghum were jointly screened by agricultural departments and scientific research institutes in Wudi County, Binzhou, Shandong Province. This year, nearly 20,000 mu of sorghum was planted, which not only ushered in a bumper harvest, but also comprehensively managed saline-alkali land. In order to cope with the high temperature weather, Tongcheng, Anhui Province has guided farmers to plant new products with high temperature resistance, and the local 630,000 mu of mid-season rice has fully entered the harvest period.

China successfully launched eight satellites, including the Celestial Instrument 41.

At 10: 31 on September 24th, China’s Taiyuan Satellite Launch Center used the Jie dragon three carrier rocket in the sea area near Haiyang, Shandong Province, and successfully launched eight satellites, namely Tianyi 41, Xingshidai -15/21/22, Yuxing-2 05, Fudan-1, Tianyan 15 and Jitianxing A-01, into the scheduled orbit.

China has built a national drug intelligent supervision platform.

Recently, the reporter learned from National Medical Products Administration that China has built a national drug intelligent supervision platform. The regulatory authorities have strengthened the application of big data in drug supervision, and promoted key varieties to achieve product line coding and traceability of the whole process. At the same time, implement a unique identification system for medical devices; Combined with the actual growth of online sales of drugs, medical devices and cosmetics, we will continue to promote the construction of online sales detection platform.

Baidu’s end-to-end speech language model is broken, the dialogue is super realistic, the cost is reduced by 90% at the highest, and the response is extremely fast in one second.

Author | Cheng Qian

Edit | Desert Shadow

Voice interaction has been played with new tricks? !

Wisdom reported on April 2nd that on Monday, Baidu released the industry’s first end-to-end speech language model based on new Cross-Attention, which has been launched online for free.

Summarize the newly upgraded essays with several key words: voice interaction.Ultra-realistic, ultra-low delay and ultra-low cost.

Super-fidelity refers to the fact that Wen Xiaoyan, equipped with a large phonetic language model, can understand Chongqing, Guangxi, Henan, Guangdong, Shandong and so onCharacteristic dialect can also realize emotional dialogue.; Ultra-low latency means that the waiting time of users can be reduced from the common 3-5 seconds in the industry toAbout 1 second, almost the same as talking to real people; Ultra-low cost means that in the voice question-and-answer scenario of telephone voice channel, the call cost is about lower than the industry average.50%-90%。

Jia Lei, chief architect of Baidu Voice, revealed that the model can be deployed on L20 card, and under the condition of meeting the delay requirement of voice interaction,The concurrency of dual L20 cards can be more than several hundred.. At present, the training process of speech language model is convenient, which can be realized by optimizing hundreds of cards in one week based on Wenxin model, and the optimization work itself is not complicated.

Compared with the application of big model in voice interaction scene, what is the uniqueness of this big model of voice language? How to reduce the call cost to 90% at the highest? How to interpret the innovation behind it? Zhi Dongxi had an in-depth exchange with Jia Lei, chief architect of Baidu Voice, trying to find the answers to these questions.

First, the real-life dialogue experience, the upgraded version of the text Xiaoyan voice interaction is more silky

The development of large-scale model in voice interaction scene is evolving towards a more natural, low-delay and highly realistic voice interaction experience. And this more anthropomorphic interactive experience, we get a glimpse of the embryonic form in the newly upgraded Wen Xiaoyan, which is equipped with an end-to-end phonetic language model and has become an emotional companion and all-round assistant.

First of all, Wen Xiaoyan has integrated information query including weather query, calendar query, unit conversion, stock price and so on.38 vertical assistantsIt can be seen that the efficiency of voice interaction is much higher than that of text interaction in these special scenarios.

Secondly, Wen Xiaoyan is rightProblems of timeliness and non-timelinessCan cope with encyclopedic inquiry, timely question and answer content of current political knowledge, and Wen Xiaoyan can search in real time, and follow the instructions accurately to reduce hallucinations; Non-timeliness issues such as common sense questions and answers are not a problem.

Finally, it is also the biggest difference between voice interaction and text. Wen Xiaoyan can be carried out with users.Emotional and natural communication, and can respond to feedback quickly, and realize realistic and anthropomorphic interaction effect.

Not much to say, let’s take a look at the actual effect of Wen Xiaoyan.

A major difficulty in speech recognition is dialect recognition. The pronunciation characteristics of dialects are rich and varied. The same dialect may have different pronunciations in different regions, and even the same word may have different pronunciations in different contexts. This makes it difficult for speech recognition system to accurately capture and analyze all pronunciation variants, which increases the difficulty of recognition. At present, Wen Xiaoyan has been able to deal with the characteristic dialects in Chongqing, Guangxi, Henan, Guangdong and Shandong. He can not only understand them, but also reply with the corresponding dialects.

Another major feature of voice communication is the need for multiple rounds of interaction. For example, in the following example, Wen Xiaoyan not only gives a method to distinguish budgerigars from different dimensions of adulthood and childhood, but also gives correct feedback in time when users interrupt to ask new questions.

Even if the reply contains many elements such as distinguishing the budgerigar’s sex and distinguishing from a certain fixed feature, Wen Xiaoyan gives a concise answer, and at the end reminds the user that he can record while observing.

In addition, human-computer interaction is often interrupted in the middle, such as the user gets the core information he wants or is not satisfied with the current output content, etc. When the user interrupts the voice playback to input, the voice recognition system may be mistaken because of environmental noise, unclear pronunciation of the user or confusion with the previous voice content.

In the face of children’s repeated interruptions, Wen Xiaoyan can accurately identify their needs of "changing a story", and when children say "Mom has already told it", they don’t automatically choose to change the story, but give an emotional reply in time to create an atmosphere of natural dialogue.

Wen Xiaoyan, an emotional interactive form, also extends its application to companion scenes in assistant scenes such as knowledge quiz. When the user mentioned "I’m in a bad mood", Wen Xiaoyan’s voice was worried and so on, and guided the user to tell the reason why he was in a bad mood and further enlighten him.

Second, brand-new cross-related attention, to create a very low training cost advantage

Different from the language model, the core difference of the phonetic language model is that it can produce emotion.

Jia Lei said that the text model only produces words, while the phonetic language model can have emotions. The key is two special links in the architectural diagram of the phonetic language model.TN Rhythm and Emotion Control of Personality and StyleThis is prepared for speech synthesis, which allows the big model to have the emotion of adapting the content while generating the answer, which is also the key innovation of Baidu’s end-to-end speech model.

Specifically, there are four key innovations.

First of all, this is released by Baidu.The industry’s first cross-modal speech language model based on Cross-Attention; The second is the model.Combining Encoder with speech recognition, the calculation of KV is reduced to 1/10.; The third isEncoder is combined with speech synthesis, and the output content can be controlled emotionally.; Finally,Efficient full query attention EALLQA, so that the KV cache is reduced to a few tenths.

On this basis, the model realizesIntegration of text recognition and text synthesis.On the basis of systematic end-to-end communication, these coupled technologies enable the model to realize a natural, realistic and emotional interactive experience on the basis of quick question and answer and quick understanding.

Jia Lei explained that the acoustic model is also a phonetic model, but usually large language models are connected by words. Therefore, in the process of integrating speech recognition and large language model, researchers share the process of Encoder and speech recognition in large language model to achieve the purpose of reducing the hard delay of speech interaction, and innovatively introduce cross-modal modeling, switching from Self-Attention to Cross-Attention, thus completing the integration of speech recognition and large language model.

Baidu proposes to use Cross-Attention to solve the cross-modal problem of speech and language. In this process, due to the speed limitation of the existing Attention technology in Cross-Attention speech language modeling, Baidu developed the EALLQA technology suitable for Cross-Attention, adopting implicit RNN two-level position coding, training MHA in 128 space, and inferring MQA in 512 space shared by all layers of the model, so as to achieveMake full use of limited training resources and reduce reasoning cost..

In the basic training of the model, Baidu uses self-distillation post-train the end-to-end speech language model of Cross-Attention based on the pre-training model of Self-Attention mature Wenxin language.

In fact,In speech model, the pressure of KV cache and KV calculation is much greater than that of text model.. Jia Lei explained that the essential difference between speech recognition and text model is that the first token at the beginning of a sentence determines the delay of speech recognition. For the large text model, users can wait for 2-3 seconds to give an answer after entering a paragraph of text, while in the large speech language model, users are less tolerant of reply delay, and they hope to hear the answer within 0.5-1 second.

On this basis, the end-to-end speech language model realizes low-cost training and low-cost high-speed reasoning. In addition, the speech language model also needs quick response and emotional response, which is the scene where another key technology is exerted.Streaming word-for-word LLM-driven multi-emotional speech synthesis. Jia Lei said that many rounds of continuous communication with emotion can make people want to continue communication.

Based on the streaming word-for-word method, its speech synthesis is to see one word jumping one word at a time, and the large model can help the speech synthesis to output the text normalized output, prosodic pause output and emotional output it needs, making the speech synthesis process flow like when people speak, and its adaptive emotional coverage can reach 17 kinds according to the text output.

In addition, there is a big pain point in speech recognition, which is that it can’t judge the starting point and ending point of the user’s speech, butLarge model blessing can make it analyze whether the user’s words have ended based on semantics.The semantics are incomplete and we need to wait.

Jia Lei further explained that,The activation of voice scenes requires extremely interactive cost, extremely fast delivery speed, intelligent and emotional humanized question and answer.. Baidu integrates speech recognition with large model to solve the problems of pre-fetching, hesitation, content understanding and quick question and answer, and integrates text synthesis with large model to output prosodic emotion needed in speech summation, thus solving the problems of context understanding and emotion control in synthesis. This makes the application potential of voice scene greatly improved.

Third, directly attack the difficulty of voice interaction, and enlarge Baidu’s end-to-end voice language model.

The continuous optimization of large model has significantly improved the robustness, naturalness and speaker similarity of speech, butThere are still many pain points in the previous technical path.This is why Baidu focuses on the end-to-end speech language model.

Compared with the communication between people, the response speed of large language model is slow, and users need to wait for a while to get a reply. In addition, voice communication is often accompanied by multiple rounds of dialogue and interaction, and it is extremely difficult for models to complete colloquial multiple rounds of interaction. Compared with texts, users use more scenes of voice interaction, and the surge of interaction will lead to the increase in the application cost of large-scale models, and the difficulty of large-scale application popularization will also increase.

On the traditional voice interaction route, it will be limited by the accurate response between contextual memory, noisy occasions, hesitant questions and interruptions.

Therefore, this has become a core contradiction in the field of voice interaction.The convenience of voice interaction determines its potential for large-scale application, and these pain points are hindering its popularization.Jia Lei believes that,The chemical reaction between speech and text is the key to find a breakthrough in a specific field in the future..

The appearance of phonetic language model is qualitative change, and its innovative synthesis technology makes the model not need to see the whole text of a sentence, but can synthesize a word by seeing the text of a word. On this basis, Baidu has excavated a unique application scenario. He gave an example. For example, when asking about the weather, users can quickly interrupt the next question when they get the temperature range of the weather. Its advantage is that the use cost of the model is greatly reduced, while the text model needs powerful hardware to achieve such efficient application, but the speech language model can achieve efficient concurrency with low-cost hardware.

At the same time, from the perspective of the whole voice interaction field, the accuracy of the speech recognition part of the large model has been greatly improved. Jia Lei believes that it is more a competition of speed, cost and answer accuracy.Cost reduction is the key to large-scale use of multimodal voice interaction..

Jia Lei said: "Cost reduction is the inevitable way of technological progress." The extremely low cost of Baidu’s speech language model means the possibility of large-scale industrialization. The landing application of AI is the core of the development of the model industry in 2025, and this model is the key to solving the speech problem.

Baidu’s accumulation in the field of speech recognition has a long history.

In 2018, the Deep Peak 2 model released by Baidu Voice broke through the traditional model that has been used for more than ten years, greatly improving the recognition accuracy in various scenarios. At the beginning of 2019, Baidu voice technology team announced the world’s first streaming multi-level truncated attention model SMLTA in the field of online voice, with a relative accuracy rate of 15%. In 2021, Baidu released the streaming truncated conformer modeling technology based on historical information abstraction-SMLTA 2, which solved the problems faced when the Transformer model was applied to online speech recognition tasks.

These technological innovations have been applied in many fields such as automobiles, consumer electronics and mobile phones. This time, in order to promote the large-scale application of speech language model, at present,Baidu has launched its online Xiaoyan and opened it for free, and plans to launch it on the open platform in April, and then it will be connected to call centers, speakers and other business lines..

Jia Lei mentioned: "Science may have national boundaries, but it has no corporate boundaries." Subsequently, Baidu opened the end-to-end speech language model, just to promote the application of the big language model in the speech field, which is conducive to the development of the whole industry and ecology.

Conclusion: Baidu’s phonetic language model is unsheathed, opening a new chapter with low cost and high efficiency.

Convenient, efficient, natural and friendly voice interaction, which is widely used in multiple scenes, is very important in the digital age. However, from the current actual effect, voice interaction has greatly improved in recognition accuracy, and now the competition focus of the industry is on its recognition speed, cost and accuracy of answers.

In this context, the release of Baidu’s end-to-end speech language model further broke through the cost, and proposed a new technical path, which pushed the industry competition of voice interaction to a new height. At the same time, Baidu will launch it on an open platform, which will accelerate the application and popularization of large models in voice interaction scenarios.

Israeli air strikes killed several senior commanders in Hamas.

  Xinhua News Agency, Gaza, May 12 (Reporter Xiong Sihao) Palestinian Islamic Resistance Movement (Hamas) confirmed on the 12th that several senior commanders of the Kasang Brigade, an armed faction affiliated to Hamas, died in Israeli air strikes on the Gaza Strip in the early hours of that morning.

  On May 12th, a building in Gaza City collapsed in an Israeli air raid, and smoke rose from the scene. Xinhua News Agency (photo by Rizek Abdel-Jawad)

  Hamas said in a statement that the Israeli army launched a fierce air raid on the Gaza strip early that morning, which killed several senior commanders of the Kasang Brigade, including the commander of the Kasang Brigade in Gaza, basem Issa.

  According to Israeli sources, the IDF and the General Administration of National Security jointly carried out targeted clearance operations in Gaza and Khan Younis in the south on the same day, killing the commanders of the Kasang Brigade in Gaza and Khan Younis, as well as 16 senior commanders in charge of network, intelligence, security and rocket engineering.

  Around midnight on May 12th, a building in Gaza City exploded in an Israeli air raid. Xinhua News Agency (photo by Yasser)

  From May 10th to 12th, a serious conflict broke out between Israel and Gaza. Armed men in the Gaza fired more than 1,000 rockets at Israel, killing six people and injuring more than 100 others. The Israeli army immediately retaliated and attacked hundreds of targets of armed groups in the Gaza. Ashraf Zidra, spokesman of the health department in the Gaza, said that 65 people were killed and 365 injured by Israeli air strikes from 10th to 12th, including 16 children.

  Since Ramadan, conflicts between Palestine and Israel have continued to occur, especially since last weekend, and tensions have intensified. On May 10th, serious clashes broke out again between Palestinians and Israeli police at the Temple Mount (called "Noble Forbidden Area" by Muslims) in the old city of Jerusalem, and more than 300 people were injured.

  On May 12th, Palestinian police inspected the damage caused by Israeli air strikes in Khan Younis, south of the Gaza Strip. Xinhua News Agency (photo by Halid Omar)

  On May 12th, Palestinians inspected the damage caused by Israeli air strikes in Khan Younis, south of the Gaza Strip. Xinhua News Agency (photo by Halid Omar)

Dongfeng Warrior 917 officially revised and pre-sold 768,000! Carbon fiber reinforced kit

Dongfeng Warriors announced that the pre-sale prices of 917 Jiaolong Armor and Jiaolong Armor Max were 768,000 yuan and 868,000 yuan respectively, and the starting price was 130,300 yuan more expensive than ordinary extended-range models. A large number of aviation-grade carbon fiber materials are added to the body of the new car, and the carbon fiber coverage rate of the whole car reaches 22.3%. The body size is larger than that of the ordinary version. It is painted in desert storm yellow (other car colors are optional) and equipped with a silver contrast roof, which is more domineering than the tank 800.

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 1

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 1

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 2

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 3

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 4

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 5

Dongfeng Warrior 917 officially revised and pre-sold 768,000 carbon fiber-reinforced kits-Figure 1

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 1

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 2

The Warrior 917 Dragon Armor was replaced by a straight waterfall type middle net, the air inlet area was increased, and two ridges were added to the machine cover. The front and rear wheel eyebrows were made of carbon fiber, and the round cake type closed wheel hub was replaced. The car side continues to use traditional door handles, adding side boxes and storage racks, and the roof is equipped with an expansion platform. Split taillights are used AT the tail, and a spare tire rack is added to place a full-size AT spare tire.

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 6

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 7

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 8

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 9

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 10

Warrior 917 Dragon Armor continues to be equipped with 12.3-inch suspended LCD instrument +15.6-inch suspended touch control screen +12.3-inch passenger entertainment screen, equipped with 4 multi-function steering wheels, streaming rearview mirrors, leather stitching seats, physical function buttons and traditional air conditioning air outlet shapes, and the central armrest box, door panel and center console adopt a large number of carbon fiber cover plates.

Add x.x million to buy the modified version of Dongfeng Warrior 917, and pre-sell it for xx.xx million yuan-Figure 1

Warrior 917 Jiaolong Armor Range Extender has a maximum power of 145kW, a total motor power of 600kW, and a peak torque of 1050N·m, matching with Contemporary Amperex Technology Co., Limited ternary lithium battery pack of 65.8kWh. CLTC has a pure electric cruising range of 200km and a comprehensive cruising range of 1026km. The new car comes standard with T-MAX professional electric winch, which can provide 12,500 pounds of pulling force.

Lei Jun: Xiaomi SU7 is positioned as a C-class high-performance eco-technology car, and the pricing is "a bit expensive".

On December 26th, Xiaomi Automobile Technology Conference will be held at 2pm on December 28th. Lei Jun, founder, chairman and CEO of Xiaomi, answered some questions that netizens are most concerned about today.

Xiaomi’s first car was named SU7. As for how to pronounce it, Xiaomi has also discussed it seriously. In the end, everyone thinks that it is still called "",just like calling a friend’s name.

In addition, Xiaomi SU7 is positioned. As for why he made a car, Lei Jun will introduce it in detail at the press conference. The press conference will start at 2 pm, which is estimated to be about 3 hours.

Lei Jun also said that Xiaomi SU7 has no accurate benchmarking vehicle, and it is in the trial production and climbing stage, and it will take several months for it to be officially listed. In terms of pricing,

The full text of Lei Jun’s answer is attached to IT House: