当前位置:网站首页>Special episode of Goddess Festival | exclusive interview with Chinese AI goddess Zhang Qingqing's transformation from a female learning tyrant to a female entrepreneur
Special episode of Goddess Festival | exclusive interview with Chinese AI goddess Zhang Qingqing's transformation from a female learning tyrant to a female entrepreneur
2022-07-25 06:10:00 【Magic Data】

Another year “ Goddess Festival ”, Nowadays, women make great contributions to society and family , Created a lot of social public value , For this reason “ Goddess power ” The word" , Express a unique power of women .
On special days , The editor interviewed the founder of AI Shu intelligence and CEO Zhang Qingqing , Let's enjoy how a female Xueba transforms into a successful female entrepreneur step by step .
To study : The indissoluble bond between female Xueba and pronunciation
Zhang Qingqing is a voice data scientist , I didn't see her for the first time “ Data scientist ” That seriousness , Instead, it gives people a capable and sunny style .
Because of a course , She is very interested in pronunciation ; Because of interest , She has been working in the field of voice ; Also because of my experience in the French National Laboratory , Let her realize the standard of good data and maintain a scientific and rigorous attitude in research .
The following is an exclusive interview with Zhang Qingqing :
Speaking of voice data scientists , It seems that women are rarely seen in this field .
Zhang Qingqing : Yes , Others say I'm a voice data scientist , But I prefer to call myself a programmer , Because at the beginning of my work, I came into contact with writing programs , One difference is that I write AI programs . I don't think there should be gender differences in any industry , Because our ideals are gender neutral .
What was the reason why you first came into contact with voice technology as a research direction ?
Zhang Qingqing : I study communication engineering myself , The first time I came into contact with pronunciation was in my junior year , At that time, I took a course called “ Speech signal digital processing ”. At that time, there were not many speech recognition experts , This course was given to us by the teacher of Datang Telecom who mainly studies communication .
Since then, I have been very interested in voice , I have no idea how people pass through their vocal cords , Then the mouth makes a sound , And I can receive sound through my ears and have a lot of interest , I hope I can do some further research in this direction . So at that time, during the doctor's escort , I chose the Institute of acoustics of the Chinese Academy of Sciences , Since then , I have been moving forward in the direction of speech recognition .
After completing the Ph.D. degree in speech signal digital processing in the Institute of acoustics, Chinese Academy of Sciences , I went to France again , Continue to engage in corresponding research work in the French National Laboratory , I finished my post doctoral study there . My research topic at that time was acoustic modeling of large vocabulary continuous speech recognition . When I was in France, I also participated in some EU voice projects , It involves many multilingual acoustic construction technologies in the European Union .
Experience in French National Laboratory , What impact does it have on your professional research in speech ?
Zhang Qingqing : At that time, I participated in an international evaluation competition in France , This is a global evaluation competition , The data used by contestants must be completely consistent . If any contestants use their own special data , Will be taken out for all contestants to share , The main thing is to ensure that all contestants evaluate their system indicators on the same level of data .
This incident illustrates one thing from the side , That is, the amount of data has a great impact on the performance of our model . Our country now also carries out some evaluation and evaluation in many industries , I also hope that these evaluations in our country can refer to overseas practices , We should evaluate it on the same scale , Make sure everyone uses the same data , Only in the end can we reflect the performance differences of each company . This can reflect our scientific and rigorous attitude , Such a rigorous attitude , It also has a great impact on my future professional research .

chart : Zhang Qingqing
Start a business : Establish AI Shu wisdom and become a female entrepreneur
With the in-depth understanding of voice technology and industry , Zhang Qingqing wants to do more and more , Finally in the 2016 In, he founded the data service company love digital intelligence . As a woman , She admitted that entrepreneurship is difficult , I also admit that I was very confident , Because AI Shu intelligence is solving the pain points of some industries .
thus , Zhang Qingqing changed from a female student bully to a successful female entrepreneur .
After returning home , What did you want to do most at that time ?
Zhang Qingqing : After returning home, I did research work in the Institute of acoustics of the Chinese Academy of Sciences , What I wanted to do most at that time was to apply my research experience of speech recognition accumulated in the past ten years to actual production and life , Contact includes Huawei 、 tencent 、 Ali 、360、UCweb Other enterprise projects , Help build the basic system of speech recognition .
In the process of doing this, I found that the whole industry lacks awareness of data , What kind of data can solve our model , Let our AI have better effect . This is lack of industry standards , So from then on , Begin to have a strong interest in data , Slowly come up with an idea , Maybe one day I will be a data company .
At that time, whether there was confidence in the establishment of the company ?
Zhang Qingqing : Of course , From the first day of entrepreneurship, we have confidence , And confidence is more and more sufficient . Probably 2015~2016 It was around the year that AI and big data surged , Many companies are developing corresponding AI software and hardware products , In the process, we found , Who uses more data 、 Better quality , The performance improvement of the product is more intuitive .
I have accumulated knowledge about how to clean in the field of artificial intelligence for many years 、 structured 、 Methods of filtering and processing data , And form a complete system , This system just solves the difficulties of artificial intelligence data processing , This point has created a blue ocean in the field of artificial intelligence .MagicData, Love number wisdom is conceived based on this point . In these five years , With the development of AI, AI is getting bigger and bigger , Recognized by more and more industry customers , Our confidence is more and more sufficient .
Especially for women , Entrepreneurship is not an easy thing .
Zhang Qingqing : First , Entrepreneurship is not an easy thing . Someone once said , Entrepreneurship is anti human , In the five years since I started my business, I feel this sentence more and more deeply .
Remember 2016 When I first started my business in, I was understaffed , Ensure the operation of the company , Many things are as small as cleaning 、 Assemble the computer , To the formation of the team 、 The operation of the company should be done personally . I remember we had a big project at that time , More than a dozen people in the company are out , Even the company's finance is helping us with data .
Is it difficult to start a business ? I think it's hard , But in this difficult process , Feel that all our colleagues are with us , Grow with the company . This love , This warmth has achieved something with everyone's efforts , It is also a very pleasant and exciting thing .
Now, what pain points does love digital intelligence solve for the industry ? How to solve these pain points ?
Zhang Qingqing : For the industry , The biggest pain point is the lack of a large number of high-quality 、 Reliable voice data , Because the quantity and quality of data determine the upper limit of machine learning . about AI For enterprises and even the whole industry , Obtaining efficient training data is the key to success . Take me , My doctoral thesis was to study the pronunciation characteristics of Chinese speaking English , Collection required 200 The voice of Chinese speaking English was used in the experiment . In order to collect this 200 Personal voice , I hired a colleague 、 friend 、 Family members and even strangers on the road , I will hold him and say , Can you record it for me ?
I spent it. 300 Days finally collected 200 Personal voice , Then handle and clean , After being sent to the system ,1 The time system of days will send it to me 1 The results of experiments . in other words ,301 In a matter of days ,300 I've been processing data all day , I'm not the only one in this situation , The whole AI industry , quite a lot AI Practitioners are facing the same problem .
Actually , We hope to spend more time on AI Optimization of the model , And how to build a better system to get a better result , But we have to stop because of the data . So today, , We love to count what wisdom hopes to do , It is to provide more professional data services for all AI workers , Let everyone no longer bother with data collection and data cleaning .
Yes AI Company speaking , Generally, there are high requirements for the amount of data , There will also be high expectations for the acquisition cycle , Based on this , Our company has stored training data for different industries and some vertical scenarios in advance . At present, we develop data according to scenarios , Including smart appliances 、 call center 、 Online education, etc , We all have corresponding training data reserves .
In the near future, we will also build a global open source community , We will put a lot of data we have stored on the open source platform , For workers who are training in artificial intelligence . We very much hope that through such an open source action , Be able to push AI Move forward faster .

chart : Zhangqingqing attends 2020 Guangzhou home appliance summit in
Open up : Promote the development of forging industry AI The goddess
in fact , Zhang Qingqing is not just doing her own business , She also takes on the responsibility of being an entrepreneur , Continue to voice for the popularization and promotion of the data industry , It also offers suggestions for the establishment of relevant specifications and Standards , Even actively participate in the formulation of industry standards .
On this special day ,AI The goddess Zhang Qingqing combines her own experience , What blessings will be brought to the vast number of female compatriots ?
As an entrepreneur , What are your visions and expectations for the voice data industry ?
Zhang Qingqing : From the first day I founded love counting wisdom , I'm very eager to bring more value to the AI industry , Push AI The industry is moving forward faster .
The data industry is an emerging industry , Many people don't know much about , It needs to be popularized and promoted repeatedly . Now love digital intelligence, including myself, hopes to have more opportunities to stand up and popularize the application of data , We also hope to participate in more industry summits and domestic and foreign exhibitions , Let more enterprises undergoing digital transformation realize the importance of data , And help them understand how to process data faster and better , It can really help them reduce costs and increase efficiency , Finally, it will bring more contributions to the industry and society .
Data as an emerging industry , In addition to the value of universal data , What else is worth doing ?
Zhang Qingqing : Data is an emerging industry , At the technical level, there are many standards that need to be regulated and formulated , Formulating normative standards is a work that will benefit the future . As early as the beginning of the company , We participated in the national standard committee 《 Chinese speech recognition system evaluation specification 》 Work out .
Last year, we discussed with industry experts , It is mainly the experts from home appliance manufacturers who discuss the technical specification issues of smart home speech recognition related standards , We hope to further promote the formulation of voice related industry standards and norms . We are also actively participating in other industries , We hope that the knowledge accumulated in the data industry , The accumulated experience , Can become a standard , Be more AI Used by practitioners .
On this special day of Goddess' day , As a successful female entrepreneur , Have you sent some messages and wishes to the majority of female compatriots ?
Zhang Qingqing : Women play a very important role in society , Women represent tenderness , At the same time, he is very resolute . I have done nearly in the field of speech recognition myself 20 year , Actually 20 Years ago, even 10 Years ago , Voice is not a very popular industry , At that time, many people didn't find their ideal job after graduation , Some may have changed careers , But interest makes me willing to continue to study in this industry . Until later, we saw the popularity of artificial intelligence , Everyone began to talk about voice 、 More and more attention is paid to images . this 20 I haven't left voice for years , I always regard it as a lifelong career .
Today is goddess day , The world is rich and colorful because of us , We always create value for our family and society 、 Wealth , But we should also remember : Be yourself ! We have the right to be ideal 、 Strive for a better life , Insist on being yourself ! These are what I want to share with you today , Thank you very much ! thank you !
Zhang Qingqing said :“ Women play many roles in society , It's a daughter , It's the mother , It's a wife , It's a colleague , It's the boss . Please don't forget , We're still on our own .”
Today is goddess day , Love counting wisdom. I wish the goddesses a happy holiday !
边栏推荐
- Function template learning record
- Android interview question: why do activities rebuild ViewModel and still exist—— Jetpack series (3)
- 有什么能在网上挣钱的项目啊?做自媒体靠谱吗?
- leetcode/ 前 n 个数字二进制中 1 的个数
- 剑指 Offer 36. 二叉搜索树与双向链表
- Vim配置Golang开发环境
- (14) [driver development] configuration environment vs2019 + wdk10 write XP driver
- 日期(DAY 76)
- Prometheus operator configures promethesrule alarm rules
- (Niuke multi school I in 2022) i-chiitoitsu (expected DP)
猜你喜欢
![[QT] solve the problem of Chinese garbled code output from QT console](/img/09/8af91d2a0327bd1d3c7b64f2b8185f.png)
[QT] solve the problem of Chinese garbled code output from QT console

【每日一练】day(14)

(2022 Niuke multi School II) l-link with level editor I (dynamic planning)

Data too long for column ‘data‘ at row 1以及设置成longblob造成的乱码解决。node-mysql

Sword finger offer 36. binary search tree and bidirectional linked list

Ffmpeg notes (I) fundamentals of audio and video
![(14)[驱动开发]配置环境 VS2019 + WDK10 写 xp驱动](/img/90/0d94d26be8128d77de65919763fda5.png)
(14)[驱动开发]配置环境 VS2019 + WDK10 写 xp驱动

HTB-Arctic

(2022 Niuke multi School II) k-link with bracket sequence I (dynamic planning)

剑指 Offer 54. 二叉搜索树的第k大节点
随机推荐
Draw Bezier curve through screen interaction
Run length test of R language: use the runs.test function to perform run length test on binary sequence data (check whether the sequence is random)
Binary search tree (day 75)
What does PK, NN, Qu, B, UN, ZF, AI, G mean when creating tables in MySQL
MySQL中建表时 pk、nn、qu、b、un、zf、ai、g代表的意思
Singing "Seven Mile fragrance" askew -- pay tribute to Jay
Softing pnGate系列网关:将PROFIBUS总线集成到PROFINET网络
ARM裸板调试之JTAG调试源码级调试
暑期总结2
A little experience about von Mises distribution
Req.body in node.express is always undefind
JS how to delete an element without deleting its child elements
Mechanism and principle of multihead attention and masked attention
Detailed explanation of arm instruction CMP
“蔚来杯“2022牛客暑期多校训练营2 Link with Game Glitch (spfa找正负环)
sqlilabs less-28~less-8a
Netease game Flink SQL platform practice
R language uses wilcox.test function to perform Wilcox signed rank test to obtain confidence interval of population median (set conf.level parameter to specify confidence level and size of confidence
[Luogu p6629] string (runs) (tree array)
y76.第四章 Prometheus大厂监控体系及实战 -- prometheus进阶(七)