DeepL

As the world becomes increasingly interconnected, the ability to communicate across languages has only become more essential and the demand for high-quality language translation has never been higher. Organizations increasingly recognize the need to communicate across borders and cultures. Businesses have to communicate with their clients, suppliers, and partners in different countries, and these communications must be accurately translated to ensure that the intended message is conveyed.
随着世界日益紧密相连，跨语言沟通的能力变得更加重要，对高质量语言翻译的需求也空前增加。组织越来越意识到跨越国界和文化进行沟通的必要性。企业必须与不同国家的客户、供应商和合作伙伴进行沟通，这些沟通必须得到准确翻译，以确保传达出预期的信息。

A company that can communicate effectively in the language of its customers or partners will be better positioned to succeed in the global marketplace. Language translation can help companies to reach new markets, expand their customer base, and establish themselves as global players in their industry The global language services market was valued at $60.7 billion in 2022 and is projected to reach $96.2 billion by the end of 2032. Advancements in technology have made language translation more accessible, easier, and cheaper than ever before, and the global language translation software market in particular was valued at $9.3 billion in 2021 and is projected to reach $44.8 billion by 2031.
一家能够有效沟通与客户或合作伙伴的语言的公司将更有可能在全球市场取得成功。语言翻译可以帮助公司开拓新市场，扩大客户群，并在其行业中确立全球参与者的地位。全球语言服务市场在 2022 年价值为 607 亿美元，预计到 2032 年底将达到 962 亿美元。技术的进步使语言翻译比以往任何时候都更易获得、更简便、更便宜，尤其是全球语言翻译软件市场在 2021 年价值为 93 亿美元，预计到 2031 年将达到 448 亿美元。

DeepL is a machine translation platform that instantly translates written content accurately and effortlessly into different languages. DeepL's advanced neural networks capture linguistic nuances and deliver real-time translations that sound human. Over 20K businesses and 1 billion users globally use DeepL through desktop and mobile apps, a browser extension, API integration, or the DeepL website.
DeepL 是一个机器翻译平台，可以将书面内容准确、轻松地瞬间翻译成不同语言。DeepL 的先进神经网络捕捉语言细微差别，并提供听起来像人类的实时翻译。全球超过 20,000 家企业和 10 亿用户通过桌面和移动应用、浏览器扩展、API 集成或 DeepL 网站使用 DeepL。

Although DeepL officially became a company in late 2017, the foundational product and original company were started 10 years earlier. In 2009, former Google research scientist Gereon Frahling founded one of the best multi-language dictionaries on the market called Linguee. It had the language pairs English-German, English-French, English-Spanish, and English-Portuguese since 2010 and provided translations for words and phrases in various languages. DeepL founder Jaroslaw Kutylowski was the CTO at Linguee from 2012 to 2017.
尽管 DeepL 在 2017 年底正式成立公司，但其基础产品和最初公司始于 10 年前。2009 年，前谷歌研究科学家 Gereon Frahling 创立了市场上最好的多语言词典之一，名为 Linguee。自 2010 年以来，它提供了英语-德语、英语-法语、英语-西班牙语和英语-葡萄牙语等语言对，并为各种语言的单词和短语提供翻译。DeepL 创始人 Jaroslaw Kutylowski 曾于 2012 年至 2017 年担任 Linguee 的首席技术官。

While trying to improve Linguee, researchers at the company started using neural networks to improve their verification process for different translations. They then started using the technology for different applications. A team led by Jaroslaw Kutylowski started training a neural network to do machine translations on the dataset they had spent the last 10 years curating, and went on to announce the launch of DeepL, an online machine translator, in 2017. In July 2019, DeepL announced the appointment of Jaroslaw Kutylowski as CEO.
在努力改进 Linguee 的过程中，公司的研究人员开始使用神经网络来改进他们对不同翻译的验证过程。然后他们开始将这项技术应用于不同的应用程序。由 Jaroslaw Kutylowski 领导的团队开始训练神经网络，以在他们过去 10 年精心策划的数据集上进行机器翻译，并宣布于 2017 年推出在线机器翻译器 DeepL。2019 年 7 月，DeepL 宣布任命 Jaroslaw Kutylowski 为首席执行官。

DeepL’s translation product uses a deep neural network architecture called a Transformer. The Transformer model is designed to overcome some of the limitations of earlier neural machine translation models (like RNNs) and has been shown to be highly effective at capturing long-range dependencies and producing fluent and natural translations. This is because RNNs process input sequences sequentially (i.e. one word at a time) and the impact of earlier words on later words diminish as the sequence gets longer. On the other hand, the Transformer model uses a self-attention mechanism that allows it to attend to all words in a sentence simultaneously, capturing both local and global dependencies.
DeepL 的翻译产品使用一种称为 Transformer 的深度神经网络架构。Transformer 模型旨在克服早期神经机器翻译模型（如 RNNs）的一些限制，并已被证明在捕捉长距离依赖关系和生成流畅自然的翻译方面非常有效。这是因为 RNNs 按顺序处理输入序列（即逐字处理），随着序列变得更长，先前单词对后续单词的影响会减弱。另一方面，Transformer 模型使用自注意机制，使其能够同时关注句子中的所有单词，捕捉局部和全局依赖关系。

DeepL supports a multitude of languages including Bulgarian, Chinese, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Indonesian, Italian, Japanese, Latvian, Lithuanian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Turkish, and Ukrainian. It has Mac, iPhone, iPad, Windows, ChromeOS, and Android applications.
DeepL 支持多种语言，包括保加利亚语、中文、捷克语、丹麦语、荷兰语、英语、爱沙尼亚语、芬兰语、法语、德语、希腊语、匈牙利语、印尼语、意大利语、日语、拉脱维亚语、立陶宛语、波兰语、葡萄牙语、罗马尼亚语、俄语、斯洛伐克语、斯洛文尼亚语、西班牙语、瑞典语、土耳其语和乌克兰语。它有 Mac、iPhone、iPad、Windows、ChromeOS 和 Android 应用程序。

Source: DeepL 来源：DeepL

DeepL Write, announced in January 2023, is an AI writing tool that improves written communication in English and German. It tackles grammar and offers suggestions on phrasing, tone, style, and word choice. DeepL Write is available on desktop and mobile browsers.
DeepL Write 于 2023 年 1 月宣布推出，是一款 AI 写作工具，可改善英语和德语的书面沟通。它处理语法并提供建议，涉及措辞、语气、风格和词语选择。DeepL Write 可在桌面和移动浏览器上使用。

Source: DeepL 来源：DeepL

DeepL enables developers to use its JSON-based REST API to embed its machine translation technology directly into their products and platforms. Companies can use its API to translate text in browsers, chat programs, email clients, or word processors. It enables the translation of Word, PowerPoint, PDF text, and HTML files. Users can choose the tone of translations for languages with formal/informal pronouns and limit the maximum monthly usage-based costs with monthly cost control. Users can also specify how words and phrases are translated by creating their Glossary. Use cases include internal communications, live translation apps, and augmented reality translations.
DeepL 使开发人员能够使用其基于 JSON 的 REST API，将其机器翻译技术直接嵌入其产品和平台中。公司可以使用其 API 在浏览器、聊天程序、电子邮件客户端或文字处理软件中翻译文本。它支持 Word、PowerPoint、PDF 文本和 HTML 文件的翻译。用户可以选择正式/非正式代词语言的翻译语气，并通过月度成本控制限制最大月度使用成本。用户还可以通过创建自己的术语表来指定单词和短语的翻译方式。使用案例包括内部沟通、实时翻译应用程序和增强现实翻译。

Translation API 翻译 API

DeepL’s customer base is split into consumers and businesses. Consumers can use the free translator tool the way they use Google Translate.
DeepL 的客户群分为消费者和企业。消费者可以像使用 Google 翻译一样使用免费的翻译工具。

On the B2B side: there are three core customer segments.
在 B2B 方面：有三个核心客户细分市场。

1. Enterprises: Companies use it to communicate with customers, partners, and employees in different languages.
   企业：公司使用它与客户、合作伙伴和员工以不同语言进行沟通。

2. Language service providers (LSPs): LSPs specialize in providing translation and interpretation services to other businesses or individuals.
   语言服务提供商（LSPs）：LSPs 专门提供翻译和口译服务给其他企业或个人。

3. Developers: Developers use DeepL's API to integrate machine translation services into their applications or software products.
   开发人员：开发人员使用 DeepL 的 API 将机器翻译服务集成到他们的应用程序或软件产品中。

The global language services industry brings in $26.6 billion a year. The global language translation software market was valued at $9.3 billion in 2021 and is projected to reach $44.8 billion by 2031.
全球语言服务行业每年收入 266 亿美元。全球语言翻译软件市场在 2021 年价值 93 亿美元，预计到 2031 年将达到 448 亿美元。

DeepL is moving away from being just a translation service and more of an AI company that focuses on text. Their latest product, Write, is aimed at taking on Grammarly, valued at $13 billion by just focusing on being a writing assistant for the English language.
DeepL 正在逐渐从仅仅是一个翻译服务转变为更注重文本的人工智能公司。他们最新的产品 Write 旨在挑战 Grammarly，后者专注于成为英语写作助手，目前估值 130 亿美元。

Unbabel - Unbabel is a language translation platform combining artificial intelligence with human translators to provide fast and accurate scaled translations. It was founded in 2013. Point 72 Ventures led their $60 million Series C to bring total funding to $91.2 million.
Unbabel - Unbabel 是一家语言翻译平台，结合人工智能和人类翻译员，提供快速准确的规模化翻译。它成立于 2013 年。Point 72 Ventures 领投了他们的 6000 万美元 C 轮融资，使总融资额达到 9120 万美元。

Lilt - Lilt is another language translation software company. It was founded in 2015 and has raised $92.5 million in funding. Lilt uses a combination of human translators and an AI translation engine. It supported 40 languages as of April 2022.
Lilt - Lilt 是另一家语言翻译软件公司。它成立于 2015 年，已融资 9250 万美元。Lilt 使用人工翻译和 AI 翻译引擎的组合。截至 2022 年 4 月，它支持 40 种语言。

Google - Google Cloud’s Cloud Translation enables companies to have their websites and applications dynamically translate text programmatically through an API. It uses Google’s Neural Machine Translation model, which Google updates on a regular cadence when more training data or better techniques become available. It can translate text for more than 100 language pairs. Its AutoML Translation lets developers create custom translation models. These custom models return results from translation queries specific to specific domains. Google Translate, the consumer-focused translation service from Google, released its Android app in January 2010, just over a year after the first commercial Android device was launched. In 2021, Google translate hit 1 billion downloads. It comes with DeepL’s free translator for consumer reach, mindshare, and usage.
Google - Google Cloud 的 Cloud Translation 使公司能够通过 API 动态地将其网站和应用程序的文本进行程序化翻译。它使用 Google 的神经机器翻译模型，当有更多的训练数据或更好的技术可用时，Google 会定期更新该模型。它可以为 100 多种语言对进行文本翻译。其 AutoML 翻译使开发人员能够创建自定义翻译模型。这些自定义模型返回特定领域的翻译查询结果。Google Translate 是 Google 提供的面向消费者的翻译服务，于 2010 年 1 月发布了其 Android 应用程序，距第一款商用 Android 设备发布仅一年多。2021 年，Google 翻译的下载量达到 10 亿次。它还配备了 DeepL 的免费翻译器，以扩大消费者覆盖范围、知名度和使用率。

Microsoft - Microsoft offers language translation through Microsoft Translator, which has both consumer and business offerings. Custom Translator is a feature of the Microsoft Translator service which enables enterprises, app developers, and language service providers to build customized neural machine translation (NMT) systems.
微软 - 微软通过 Microsoft Translator 提供语言翻译，该服务既面向消费者又面向企业。Custom Translator 是 Microsoft Translator 服务的一个功能，它使企业、应用开发人员和语言服务提供商能够构建定制的神经机器翻译（NMT）系统。

Amazon - Amazon offers Amazon Translate, a neural machine translation service that delivers fast language translation. It supports 75 languages. With Active Custom Translation (ACT), Amazon Translate allows developers to take greater control over machine-translation output by letting them bring their data to Amazon Translate to customize the machine-translated output to suit their needs. ACT produces custom-translated output without the need to build and maintain a custom translation model.
亚马逊 - 亚马逊提供亚马逊翻译，这是一项提供快速语言翻译的神经机器翻译服务。它支持 75 种语言。通过 Active Custom Translation (ACT)，亚马逊翻译允许开发人员更好地控制机器翻译输出，让他们将自己的数据带入亚马逊翻译，以定制机器翻译输出以满足其需求。ACT 生成定制翻译输出，无需构建和维护自定义翻译模型。

DeepL uses a freemium business model that funnels users into paid tiers. The free tier limits the number of characters a user can translate at once, the security of the user’s data, the number of documents they can translate a month, the number of glossaries they can have, and the integrations allowed. The free tier for the DeepL API, which has access to the same features, is less restrictive, allowing access to more glossaries and more translations.
DeepL 采用了一种免费增值业务模式，将用户引导至付费层。免费层限制了用户一次可以翻译的字符数、用户数据的安全性、用户每月可以翻译的文档数量、用户可以拥有的术语表数量以及允许的集成数量。DeepL API 的免费层对相同功能的访问不那么严格，允许访问更多的术语表和更多的翻译。

DeepL Pro is for individuals and teams and has three tiers that charge per user per year: the starter tier costs $104.88/year, advanced is $344.88/year, and ultimate is $689.88/year. They also have enterprise pricing for teams larger than 10 people, which businesses need to contact sales for.
DeepL Pro 适用于个人和团队，有三个层次，按用户每年收费：入门级每年 104.88 美元，高级每年 344.88 美元，终极每年 689.88 美元。他们还为超过 10 人的团队提供企业定价，企业需要联系销售部门。

DeepL API Pro, which offers fewer restrictions and more functionality, charges users on a monthly basis of $5.49/month or $65.88/year plus $25.00 per 1 million characters translated.
DeepL API Pro 提供更少的限制和更多的功能，按月收费 5.49 美元/月或 65.88 美元/年，另外每翻译 100 万个字符收取 25.00 美元。

Source: DeepL 来源：DeepL

Over one billion people have used DeepL’s services. In late 2021, they were the 9th most visited website in Switzerland, the 30th most visited website in Japan, and the 146th most visited site in the world. 20K businesses around the world use DeepL. It supports 31 languages, and its annual run rate was $50 million at the end of 2022.
DeepL 的服务已经被超过十亿人使用过。到了 2021 年底，他们是瑞士第 9 大访问量最高的网站，在日本是第 30 大访问量最高的网站，在全球是第 146 大访问量最高的网站。全球有 2 万家企业使用 DeepL。它支持 31 种语言，截至 2022 年底，年度运营额达到 5000 万美元。

DeepL announced it had raised over $100 million in January 2023 at a $1 billion valuation led by IVP.
DeepL 宣布在 2023 年 1 月由 IVP 领投，融资超过 1 亿美元，估值达到 10 亿美元。

For each additional language DeepL adds, it has an opportunity to upsell existing customers as well as add larger enterprises with big contracts. It’s important to do this as fast as possible because it’s hard to get large enterprises to switch their translation engines.
对于 DeepL 添加的每种额外语言，它都有机会向现有客户提供升级服务，同时吸引签署大合同的大型企业。尽快完成这一步骤非常重要，因为让大型企业更换其翻译引擎是一项艰巨的任务。

DeepL is currently coming out with its Grammarly competitor, Write. Grammarly, going after only English speakers, has attained a $13 billion valuation, with 30 million people using it a day. DeepL is poised to create the same product for each of the 31 languages it currently supports. If Write starts being adapted more globally, it could be a new source of revenue in a domain that has been less competitive than direct machine translation.
DeepL 目前正在推出其与 Grammarly 竞争的产品 Write。Grammarly 仅面向英语使用者，已经获得了 130 亿美元的估值，每天有 3000 万人使用。DeepL 准备为其目前支持的 31 种语言中的每一种创建相同的产品。如果 Write 开始在全球范围内得到更多的应用，它可能成为一个新的收入来源，在一个比直接机器翻译竞争更少的领域。

Despite millions of users, many customers and businesses they are going after are using one of Google, Microsoft, or Amazon’s cloud services. DeepL has historically performed better enough on translations for businesses to justify using DeepL over a platform like Google Cloud, even if that business already uses Google Cloud for other services. However, the moment DeepL’s edge in translation becomes less distinct, it could lose more enterprise customers because it is easier for the business to go with the offering on the platform the business is already on.
尽管有数百万用户，但许多客户和企业正在使用谷歌、微软或亚马逊的云服务。DeepL 在历史上在翻译方面表现得更好，足以让企业选择 DeepL 而不是像谷歌云这样的平台，即使该企业已经在使用谷歌云提供的其他服务。然而，一旦 DeepL 在翻译方面的优势变得不那么明显，它可能会失去更多企业客户，因为企业更容易选择已经在的平台上提供的服务。

When DeepL was getting started, its technological moat lay partially in the curated and high-quality data it got from the multilingual dictionary it spent the last couple of years on. Now, that moat is largely gone. This puts more pressure on DeepL’s ability to build the best transformer, which is one of the company’s primary technological moats.
当 DeepL 刚开始时，其技术壕沟部分在于它从多语言词典中获取的经过筛选和高质量的数据。现在，这种壕沟基本上已经消失。这给 DeepL 建立最佳转换器的能力增加了更多压力，这是公司主要的技术壕沟之一。

DeepL is an AI-powered machine translation company with the goal of making communication and translation seamless for individuals and businesses. After originally being founded as a multilingual dictionary, Linguee, the company pivoted to full machine translation after discovering how accurate translations were on their data with the new transformer technology. Producing one of the best machine translation models in a very competitive market, they now have their sights set on disrupting the multi-billion dollar AI-powered writing assistant market for each of the languages they offer.
DeepL 是一家以人工智能为动力的机器翻译公司，旨在使个人和企业之间的沟通和翻译变得无缝。最初成立为多语言词典 Linguee 后，该公司在发现新的变压器技术在其数据上进行准确翻译后，转向全面机器翻译。在竞争激烈的市场中生产出最佳的机器翻译模型之一，他们现在将目光投向颠覆每种语言提供的数十亿美元的人工智能写作助手市场。