跟讀練習: Episode 72 : Neuralk - AI for Structured Data - 通過影片學習英語口語

選擇練習模式

每種模式都會開啟全屏練習室。

161 句
如果句子過短或過長,請點選 Edit 進行調整。
1
Hey, everyone, and welcome back.
2
Today, we're going deep on a company that's tackling, well, a challenge you might not have even realized existed.
3
Yeah.
4
AI for the stuff that actually runs most businesses, structured data.
5
It's like, you know, everyone's obsessed with AI writing poems or making art, but this company is focused on the unsung hero of the data world.
6
Yeah.
7
The spreadsheets and databases we all rely on.
8
OK, so tell me more.
9
Yeah.
10
What exactly does this company do?
11
They're called Neuralc AI, and they're based in Paris, France.
12
Ooh la la.
13
Founded by Alexandre Pescu and Antoine Musnot.
14
Okay.
15
Their whole mission is to build AI specifically designed to understand and analyze structured data.
16
Okay.
17
Particularly in the commerce sector.
18
Paris, France.
19
That's kind of cool.
20
Yeah.
21
It adds a little mystique to the AI world, right?
22
It does.
23
But what's the problem they're solving?
24
Right.
25
Is analyzing spreadsheets really that hard?
26
It's not just about analyzing.
27
Okay.
28
It's about understanding.
29
Got it.
30
Most businesses are drowning in data, but they're only scratching the surface of what they can learn from it.
31
Yeah.
32
Traditional analytics can tell you what happened, but Neuroc .ai is building AI that can predict what will happen and help businesses make smarter decisions.
33
So it's like having a crystal ball for your business data.
34
Kind of, yeah.
35
Give me an example.
36
Okay.
37
What kind of problems can they solve with this AI?
38
Think about a massive online retailer with millions of products.
39
Just keeping their product catalog clean and consistent is a huge task.
40
I can imagine.
41
Imagine trying to make sure every product description is accurate, categorized correctly, and all the images are high quality.
42
It's a logistical nightmare.
43
Oh, I bet.
44
And humans probably make mistakes or things slip through the cracks.
45
Exactly.
46
That's where Neuralg AI comes in.
47
They've built models that can automatically clean up product catalogs,
48
standardize information, and even enrich existing data with additional attributes.
49
Wow.
50
They actually did this for a major European retailer called Akan Retail, and the results were impressive.
51
So instead of having a team of people manually combing through millions of product entries, they have AI doing the heavy lifting.
52
That's got to save them a ton of time and money.
53
Absolutely.
54
And it reduces errors, too.
55
And it reduces errors, too.
56
But it goes beyond just cleaning up data, right?
57
Right.
58
They're also using AI to personalize customer experiences.
59
Exactly.
60
Ooh, that's where it gets really interesting for the shoppers like me.
61
Yeah.
62
Tell me more about that.
63
By analyzing things like purchase history and browsing behavior, all structured data, by the way,
64
their AI can predict what products a customer is most likely to want and recommend them accordingly.
65
So it's like those, you might also like the gestures you see online, but powered by really sophisticated AI that understands your individual preferences.
66
Exactly.
67
And it's not just based on simple keyword matching.
68
Okay.
69
They have a model called a tabular foundation model that actually understands the underlying meaning of product attributes.
70
It's kind of like how some AI can understand images or text.
71
Right.
72
But this one is built for tabular data.
73
Okay.
74
The kind you find in spreadsheets and databases.
75
Wait, hold on.
76
Tabular foundation model.
77
That sounds pretty technical.
78
It is a little bit, yeah.
79
Do you break that down for me and the listeners?
80
Sure.
81
What makes it so special?
82
Think of it this way.
83
Okay.
84
Imagine a chef who can not only read a recipe that's like basic analytics.
85
Okay.
86
But also understand flavor profiles and predict how different ingredients will work together.
87
Yeah.
88
That's what this model can do with data.
89
I see.
90
It can see patterns and relationships that traditional methods would miss.
91
Okay, that's a great analogy.
92
Thanks.
93
So it's not just about crunching numbers.
94
Right.
95
It's about understanding the deeper meaning behind those numbers and using that knowledge to make predictions.
96
Exactly.
97
Okay, I'm starting to see how this could be a game changer for companies.
98
Yeah.
99
What other cool stuff have they done with this tabular foundation model?
100
Another example is in similarity searches.
101
Okay.
102
Imagine you're shopping online for a new pair of shoes.
103
Instead of just relying on basic keywords, this model can understand the underlying meaning of product attributes,
104
like style, color, material, and even brand reputation.
105
So it could suggest shoes that are similar in style, even if they don't share the exact same keywords in the description.
106
Yes.
107
It's about understanding the essence of what you're looking for, not just matching words.
108
Right.
109
This makes product discovery much more intuitive and helps customers find what they really want, even if they don't know the exact terms to use.
110
This is all very cool stuff.
111
Yeah.
112
And I can see how this kind of AI could be applied to so many different industries beyond just retail.
113
You're absolutely right.
114
Yeah.
115
Structured data is.. everywhere.
116
And the potential applications for this technology are vast.
117
Well, listeners, it sounds like Neurok AI is definitely a company to watch for them.
118
Where are they headed next?
119
Well, they recently secured $4 million in funding to further develop their technology, specifically their tabular foundation models.
120
That's a pretty significant investment.
121
What are they planning to do with that funding?
122
It seems they're staying laser focused on the commerce sector makes sense
123
which makes sense given their early success there yeah
124
they're also gearing up to launch the first official version of
125
their model oh cool very soon with a full benchmark release expected later this year oh a benchmark release
126
that sounds interesting what does
127
that mean for people like me who might not be you
128
know super technical think of it like a public test they'll be releasing their model
129
and data to researchers and developers so they can see how it performs compared to other AI models out there.
130
It's a way to prove the effectiveness of their technology and show the world what it can do.
131
So it's kind of like a competition for AI models.
132
In a way, yeah.
133
It's a chance for Neuralc AI to demonstrate how their approach to structured data AI stacks up against the competition.
134
This is all very exciting.
135
It sounds like Neuralc AI is really pushing the boundaries of what's possible with AI and data.
136
Absolutely.
137
They're at the forefront of a very important trend.
138
Yeah.
139
We're moving beyond the era of just collecting data and analyzing it in a basic way.
140
Now it's about using AI to truly understand the meaning behind that data and unlock its full potential.
141
It makes you wonder what the world will look like when AI can truly understand all types of data.
142
Right.
143
Not just structured data.
144
It's a fascinating question to consider.
145
Yeah.
146
Imagine a world where AI can seamlessly integrate information from spreadsheets, databases, text documents, images, even audio and video.
147
Wow.
148
The possibilities are almost endless.
149
That's pretty mind -blowing.
150
But for now, it sounds like Neuralc AI is making incredible strides in helping businesses make sense of the data they already have.
151
Exactly.
152
And their upcoming model launch and benchmark release will be key milestones to watch.
153
Right.
154
I'm very curious to see what kind of impact they'll have on the industry.
155
Me too.
156
I'm definitely adding Neuralc AI to my list of companies to keep an eye on.
157
Well, listeners, that brings us to the end of this deep dive into Neuralc AI.
158
But don't worry, the learning doesn't stop here.
159
We'll be back tomorrow evening with another exciting deep dive into a new topic.
160
Make sure you subscribe and hit that notification bell so you don't miss it.
161
Until then, keep exploring, keep learning, and keep asking those big questions.

下載應用

流利口語所需的一切

AI 發音評分逐句打分
音標練習掌握每個音
詞彙學習建立你的詞庫
詞彙遊戲邊玩邊學
FOR YOU

推薦影片

本課的詞彙與口語要點

這節口語課以影片“Episode 72 : Neuralk - AI for Structured Data”為素材。 影片中反覆出現的詞有:model, product, Neuralc, structured, exactly。 這段影片共有 161 個句子、1305 個單詞可供跟讀。 講話部分時長為 7:33。 說話人語速自然,每分鐘約 173 個詞,接近日常對話。 82% 的單詞屬於英語最常用的 3,000 詞,其餘的詞建議在練習前先學一下。

影片中的重點詞彙

影片中 15 個值得學習的單詞,附發音和釋義:

單詞發音釋義
analyze 動詞/ˈæn.əˌlaɪz/分析
database 名詞/ˈdeɪtəˌbeɪs/資料庫 /資料庫
predict 動詞/pɹɪˈdɪkt/預言 /預言
attribute 名詞/ˈæt.ɹɪˌbjut/屬性 /屬性
listener 名詞/ˈlɪs(ə)nɚ/傾聽者 /傾聽者, 聽者 /聽者
solve 動詞/sɒlv/解決 /解決
commerce 名詞/ˈkɑ.mɚs/貿易 /貿易, 交易
rely 動詞/ɹɪˈlaɪ/依賴 /依賴, 依靠
shoe 名詞/ˈʃu/鞋子, 鞋
underlying 形容詞根本, 基本
dive 動詞/daɪv/潛水 /潛水
analytics 名詞/ˌæn.əˈlɪt.ɪks/分析學 /分析學
retailer 名詞/ˈɹiteɪlɚ/零售商
topic 名詞/ˈtɑpɪk/主題 /主題, 話題 /話題
vast 形容詞/væst/浩大

需要注意的發音

說話人用了 25 次縮略和弱讀形式,例如 they're, don't, I'm。請按聽到的簡短形式來說。

  • “sh” 和 “zh” 音: benchmark /ˈbɛn(t)ʃmɑːk/, shoe /ˈʃu/, reputation /ˌɹɛpjʊˈteɪʃən/, chef /ʃɛf/, prediction /pɹɪˈdɪkʃən/
  • 長單詞——注意重音位置: analytics /ˌæn.əˈlɪt.ɪks/, reputation /ˌɹɛpjʊˈteɪʃən/, developer /dɪˈvɛləpɚ/, effectiveness /ɪˈfɛk.tɪv.nɪs/, accordingly /əˈkɔɹ.dɪŋ.li/

中文母語者容易讀錯的音:

  • /v/ — 不要讀成 /w/,上齒輕觸下唇: solve /sɒlv/, dive /daɪv/, vast /væst/, impressive /ɪmˈpɹɛsɪv/, discovery /dɪˈskʌv.ɹi/
  • 詞尾子音 — 要讀清楚,後面不要加母音: analyze /ˈæn.əˌlaɪz/, database /ˈdeɪtəˌbeɪs/, predict /pɹɪˈdɪkt/, attribute /ˈæt.ɹɪˌbjut/, benchmark /ˈbɛn(t)ʃmɑːk/
  • /r/ 和 /l/ 的區別: rely /ɹɪˈlaɪ/, retailer /ˈɹiteɪlɚ/, crystal /ˈkɹɪstəl/, explore /ɪkˈsploɹ/, correctly /kəˈɹɛk(t).li/

如何用這段影片練習

  1. 先完整聽一遍影片,不要開口,記下不認識的單詞。
  2. 先用 0.75 倍速逐句跟讀,熟練之後再回到正常速度。
  3. 錄下自己的聲音並與原聲對比,特別注意 analyze, database, predict 這類單詞。

什麼是跟讀法?

跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧,最初開發用於專業口譯員的培訓,並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大:您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同,跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性,語調,節奏,連讀,聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。

影子跟讀法: 閱讀完整分步指南 →