Written by Wu Hongxia, China Council for the Promotion of International Trade Patent and Trademark Office

“Xiao Ai, what is the weather today?” “Xiao Du, search for the recipe of braised beef with tomatoes.” Smart devices such as computers, smartphones, and smart speakers can perform corresponding operations immediately after “understanding” voice commands. Approximately 20% of the information processed by the human cerebral cortex daily is auditory information, making it one of the most important links in communication. With the continuous development of artificial intelligence technology, the experience of human-machine voice interaction is becoming increasingly popular, leading to iterative changes in people’s work and lives.

According to the “White Paper on the Development of China’s Intelligent Voice Industry” released by the National Industrial Information Security Development Research Center on December 27, 2020, the intelligent voice industry has transitioned from typical cases of essential scenarios to large-scale applications, deeply integrating into fields such as education, healthcare, government, finance, telecommunications, and justice; personal scenarios focus on home, automotive, and office environments.[1]

1. Definition of Voice Wake Words

For the purpose of discussion, this article first defines “voice wake words.” Broadly speaking, “voice wake words” refer to voice commands that can awaken smart devices to enter a working state. Smart devices match the externally monitored voice wake words with their pre-set keywords to initiate operations.

Based on the different entities that set them, voice wake words can be divided into:

▶ Pre-set voice wake words by companies (e.g., the voice wake word “Xiao Du” pre-set by Baidu in smart speakers)

▶ User-defined voice wake words.

Representative application scenarios for voice wake words include:

▶ Traditional voice interaction: first waking up the device, waiting for feedback, and then issuing voice control commands.

▶ Combined commands: directly stating the voice wake word and the operational command together, such as “Xiao Ai, what is the weather today?” The smart device will directly initiate the corresponding operation after being awakened.

▶ Unconscious wake: setting frequently used voice operational commands as wake words, allowing the smart device to directly initiate corresponding operations without the user’s conscious wake, such as saying “Navigate to work” to a smartphone.

User-defined voice wake words primarily meet personalized needs and do not establish a clear, stable connection with companies and their products. Therefore, companies do not have any rights over user-defined voice wake words, which are not within the scope of this article’s discussion. Additionally, general statements that lack distinctive features, whether pre-set by companies or user-defined on smart devices (e.g., “Navigate to work” in unconscious wake scenarios), also do not fall within the scope of this article’s discussion. General voice wake words belong to the public domain and should not be owned by any specific entity.

In summary, the voice wake words discussed in this article are limited to those pre-set by companies on smart devices that possess certain distinctive features, such as “Xiao Ai” and “Xiao Du” mentioned at the beginning of this article.

2. Legal Nature of Voice Wake Words

Innovation is the primary driving force for development, and protecting intellectual property is protecting innovation. Voice wake words present a new topic for intellectual property protection brought about by the development of new business models.

Can prior voice wake words prohibit peers from using the same or similar wake words? This requires an analysis of the legal nature of voice wake words. Currently, opinions in the industry are divided.

One viewpoint is that for actual users, the operation of waking up smart terminals using voice wake words is essentially a technical functional reference symbol, akin to a virtual switch. At this point, when users hear the wake word, they only associate it with a virtual key to unlock the smart terminal, with no other implications.[2]

Another viewpoint is that voice wake words are a form of commercial marking symbol, i.e., trademarks. Trademarks are signs used to identify and distinguish the source of goods or services. Even if voice wake words do not perform identification and distinction functions during the purchasing phase, actual or potential users can still identify the source of the corresponding smart device through voice wake words during the usage phase, thus distinguishing different smart devices. For example, “Xiao Ai” corresponds to smart devices such as smartphones, smart speakers, and smart TVs from Xiaomi; “Xiao Du” corresponds to smart speakers, visual phones, and mobile TVs from Baidu. Therefore, voice wake words reflect the essential attributes of trademarks and should be included in trademark protection.

Another viewpoint is that, like product names, company names, network domain names, and website names, voice wake words are rights protected under Article 6 of the Anti-Unfair Competition Law. Although this emerging right is not explicitly stated in Article 6 of the Anti-Unfair Competition Law, if it generates a certain impact during use, it should be included in the protective scope of the catch-all clause of Article 6, paragraph 4.

Currently, the legal nature of voice wake words remains in a stage of subjective interpretation, with no definitive conclusion. When a new phenomenon arises, people often attempt to interpret and classify it within existing institutional frameworks. When the development of a new phenomenon is substantial enough to impact social life, and the existing institutional framework cannot accommodate it, new regulations are established. It is hoped that relevant institutions will quickly provide authoritative interpretations of the legal nature of voice wake words or grant them specialized legal protection, enabling companies in China’s smart product industry to formulate and implement appropriate strategies for protecting voice wake words.

3. Voice Commands for Smart Products

First Lawsuit Case

Unlike the booming development of China’s intelligent voice industry, there are very few voice wake words or voice commands that have entered the administrative or judicial protection scope.

Through case searches, I have not found any cases protecting voice wake words or voice commands under the Trademark Law. The first judgment regarding voice commands for smart products was made by the Haidian District People’s Court of Beijing on December 31, 2020, in the case of Baidu Online Network Technology (Beijing) Co., Ltd. vs. Beijing Zile Technology Co., Ltd. and Beijing Jingwei Zhicheng E-commerce Co., Ltd. for unfair competition.[3]

In this judgment, the court granted protection to the voice command “xiaoduxiaodu,” which has a certain impact on Baidu Online, based on the catch-all clause of Article 6, paragraph 4 of the Anti-Unfair Competition Law. Regarding the legal nature of voice commands, the court stated: although voice commands in human-computer interaction are not explicitly listed in Article 6 of the Anti-Unfair Competition Law, the purpose of this provision is to prevent confusion and avoid misidentification by the relevant public. Therefore, even though voice commands are a product of artificial intelligence development at a certain stage and appeared later than product names, company names, network domain names, and website names, and there are certain differences in types, as long as they can establish a specific connection with the goods or services and their providers, and have a certain impact, they should be included in the protective scope of rights specified in Article 6 of the Anti-Unfair Competition Law.

The court found that whether waking up the Xiao Du smart speaker from standby mode or executing a specific operation, it is necessary to call “xiaoduxiaodu.” It is evident that “xiaoduxiaodu” has become an essential and frequently used specific voice command for users when using the Xiao Du smart speaker, establishing an inseparable connection with the product’s human-computer interaction and other functions and services. After extensive promotion, the voice command “xiaoduxiaodu” has established a clear and stable connection with Baidu Online and its products, possessing high recognition and significant influence. The court further determined that the defendant’s use of the voice command “xiaoduxiaodu” (written as “Xiao Du Xiao Du”) on the learning machine violated the provisions of Article 6, paragraph 4 of the Anti-Unfair Competition Law, ordering the defendant to pay 550,000 RMB in compensation.

4. Intellectual Property Protection of Voice Wake Words

China’s intellectual property legal system mainly consists of specialized laws for patents, trademarks, and copyright, along with the Anti-Unfair Competition Law. Before relevant institutions provide authoritative interpretations of the legal nature of voice wake words or grant them specialized legal protection, the protection of voice wake words should still be considered within the existing intellectual property legal framework in China.

1. Protecting voice wake words under the Anti-Unfair Competition Law. After the first-instance judgment in the aforementioned “xiaoduxiaodu” voice command unfair competition case was made, the defendant did not appeal. Currently, the promotional video on the defendant’s official website has changed the wake word from “xiaoduxiaodu” to “Du Yaya Tongxue.” The resolution of this case seems to pave a broad path for the future protection of voice wake words through the Anti-Unfair Competition Law. However, according to the Supreme People’s Court’s notice on fully leveraging the role of intellectual property trial functions to promote the great development and prosperity of socialist culture and facilitate coordinated economic development (Fa Fa [2011] No. 18), the supplementary protective role of the Anti-Unfair Competition Law must not conflict with the legislative policies of specialized intellectual property laws. In areas where specialized intellectual property laws have exhaustive provisions, the Anti-Unfair Competition Law generally does not provide additional protection, allowing free use and competition, but can still provide protection from the perspective of preventing unfair competition within the scope compatible with the legislative policies of specialized intellectual property laws. Thus, the Anti-Unfair Competition Law plays a catch-all role in China’s intellectual property legal system, and can only appropriately function after not conflicting with the provisions and legislative policies of specialized intellectual property laws while reserving space for “free use and fair competition.” Furthermore, in handling individual cases, to seek protection under Article 6 of the Anti-Unfair Competition Law, the plaintiff must provide sufficient evidence of a market competition relationship with the defendant, establish a specific connection between the voice wake word and the goods or services and their providers, demonstrate a certain impact through extensive promotion and use, and show that the defendant’s actions constitute unfair competition. Therefore, rights holders must first conduct extensive promotion and use of voice wake words to generate a certain impact before invoking the Anti-Unfair Competition Law for protection.

2. Registering voice wake words as word trademarks for protection. For example, “SIRI” is a voice assistant used by Apple Inc. on products such as iPhone, iPad, iPod Touch, and HomePod. With Siri, users can search for information, make calls, send messages, get directions, play music, and locate Apple devices. Through searches, I noticed that Apple Inc. has registered the trademark “SIRI” and “Hey Siri” in various classes, including 9, 11, 12, 14, 28, 35, 37, 38, 39, 41, 42, 43, 44, and 45. Registered trademark rights are a strong public right that is effective nationwide, and in the process of rights protection, the rights holder does not need to bear the burden of proof regarding the prior registered trademark’s recognition. Therefore, compared to the protection of rights under Article 6 of the Anti-Unfair Competition Law, registered trademark protection has inherent advantages. The difficulty in protecting voice wake word trademarks lies in determining whether they are used in the same or similar goods and services as the registered trademarks. Taking the voice wake word “SIRI” used on smartphones as an example, as mentioned earlier, Siri serves as a virtual assistant. According to Article 22 of the Trademark Law, trademark applicants must fill out the product categories and names in accordance with the standardized product classification table when applying for registration. However, “virtual assistant” is not a standardized product or service item, making it difficult to accept in trademark registration applications. In such cases, it is generally chosen to register the trademark in the product categories corresponding to the smart devices (e.g., Class 9 for smartphones, smart speakers, etc.), Class 9 for “data processing equipment, downloadable computer application software,” Class 42 for “platform operation services (PaaS),” and other related goods or services. Therefore, according to the requirements for trademark registration applications in China, it is challenging to apply for registration of voice wake words in the actual goods or services; they can only be registered in related goods and service categories. This objective situation will inevitably lead to disputes in future trademark infringement lawsuits regarding whether the accused voice wake word was used in a trademark sense on the same or similar goods and services as the prior registered trademark, becoming a focal point of contention between the parties and a challenge for the court’s examination.

3. Registering voice wake words as sound trademarks for protection. Protecting voice wake words naturally leads to the path of sound trademarks. A sound trademark is a trademark constituted by the sound itself used to distinguish the source of goods or services. Sound trademarks can consist of musical sounds (e.g., a piece of music), non-musical sounds (e.g., human or animal sounds), or a combination of both. Submitting a sound trademark application requires providing a sound sample that meets the requirements, describing the sound trademark through staff notation, simplified notation, or text. This requires the sound trademark to have certainty and stability, not changing with the usage context. Although voice wake words are designed and specified by companies, they are commands issued by users and can vary with different users. Therefore, voice wake words do not fully meet the requirements for sound trademark registration. Additionally, sound trademarks generally lack inherent distinctiveness and need to be used extensively or for a long time to establish a stable connection with the applicant, possessing the function of distinguishing the source of goods or services to acquire distinctive features. This requires companies to use voice wake words extensively or for a long time before obtaining sound trademark registration. In practice, the evidentiary burden for a sound trademark to “acquire distinctive features” is often higher than the burden of proof for “certain impact” under the Anti-Unfair Competition Law.

4. Protecting voice wake words as works under the Copyright Law. The works referred to in China’s Copyright Law are intellectual achievements that possess originality and can be expressed in a certain form within the fields of literature, art, and science. To be protected under the Copyright Law, voice wake words must meet the originality requirement. Voice wake words are generated to facilitate user operation of smart devices, often consisting of short and memorable words (generally between 2-4 characters). Although they possess certain distinctive features, they are difficult to exhibit artistic beauty and reach a certain level of intellectual creativity. Therefore, voice wake words cannot meet the originality requirement for works and are challenging to be protected under copyright.

5. Strategies and Recommendations

1. Although the legal classification of voice wake words is still inconclusive, both enterprises and the judiciary believe that granting certain legal protection to voice wake words has significant practical importance. This article calls for relevant institutions to quickly provide authoritative interpretations of the legal nature of voice wake words or grant them specialized legal protection to meet the intellectual property protection needs of companies in China’s smart product industry.

2. The product or service category is one of the central links in the entire trademark registration process, determining the scope of protection for registered trademarks. Traditional industries have stable representations of product or service categories due to long historical practices. However, new business models and fields are rapidly evolving, leading to swift changes in product and service categories. It is recommended that trademark examination institutions keep pace with the times, timely incorporating mature new product and service categories from emerging industries into the “Similar Goods and Services Classification Table” to further enhance the level and efficiency of intellectual property protection.

3. “Preparation leads to success, while lack of preparation leads to failure.” Companies in China’s smart product industry should carefully select voice wake words and formulate and implement protection strategies for voice wake words that align with their circumstances. According to industry business practices, voice wake words are often chosen to be easy to read, memorable, and aesthetically pleasing. According to the “Most Comprehensive AI Wake Word Research Report” published by Beijing Baidu Network Technology Co., Ltd. in 2018,[4] 1) the existing wake word composition method: names starting with “Xiao + character” are the most favored by users; 2) tone: users prefer the first tone (yinping), while they prefer “ping sheng” (first tone and second tone) over “ze sheng” (third tone and fourth tone); 3) initials: users prefer words with zero initial sounds, while those containing z, c, and s are least favored; 4) finals: based on the mouth shape when pronouncing finals, users prefer finals that are more open; 5) according to the structure of finals, words with single finals are the most favored by users. Given the limited resources of such words, the likelihood of different market entities coincidentally choosing the same or similar voice wake words increases. This requires relevant market entities to actively avoid using words that are the same or similar to influential voice wake words or well-known trademarks through proactive searches or market research to avoid unnecessary trademark infringement or unfair competition disputes.

Notes

[1] See: https://www.sohu.com/a/441156965_120419484, accessed on 2021.12.9.

[2] He Jingwei. Analysis of the Legal Nature of Using Others’ Trademarks as Wake Words. Journal of China Trademark, 2019(02):75-77.

[3] See the civil judgment of the Haidian District People’s Court of Beijing (2019) Jing 0108 Minchu 63253.

[4] See: https://baijiahao.baidu.com/s?id=1590351415202797724&wfr=spider&for=pc, accessed on 2021.12.9.

Source: IPRdaily Chinese Network (iprdaily.cn)

(This article is an authorized publication and represents the author’s views. Unauthorized reproduction is prohibited.)

“Star Mark,” “Forward,” “Look at,” give the editor a chicken leg!”For submissions, please contact [email protected]

Recent Activities 1

We welcome all intellectual property legal professionals to actively participate 👇

We hope legal and industry professionals in the new cultural and entertainment field will enthusiastically recommend 👇

Recent Hot Articles

• NFT and Digital Copyright Practice

• What are the “Other Relationships” between Trademark Registrants and Registrants?

• Legal Analysis of “Ghost Videos” from the Perspective of Copyright Law

• Xiong Qi: A Different Perspective on Restricting Exclusive Licenses

• Can Overlooked “Copyright Disputes” Affect Corporate Listings?

• Analysis of Patent Mining in the Context of the “Three Character Classic”

• Is Repeated Litigation Triggered by Continuous Infringement of Intellectual Property Rights Considered Repetitive Litigation?

• Can Registered Trademarks Infringing on the Rights of Influential Product Names Be Prohibited from Use?

• Breaking! USPTO Adds Administrative Sanction Procedures for Abnormal Trademark Applications

• Common Legal Issues in Exclusive Game Distribution and Operation Agreements

• Discussion on Game | Discussion on Personal Information Protection Law

• Interpretation by the Handling Lawyer of the First Case Against “Providing Game Leveling Services to Minors” Ban

• Brief Discussion on Copyright Infringement Issues in Software Reverse Engineering

• Matters Regarding Trademark Registration of Movie Titles

• Who Published the Most Intellectual Property Academic Articles in 2020?

• [Heavyweight] 2020 Annual Ranking of Institutions Publishing Intellectual Property Professional Articles!

• How Market Entities Can Avoid Risk Points When Publishing Internet Information