SoundHound AI Inc (SOUN) 2022 Q4 法說會逐字稿

Good day, and welcome to the SoundHound Q4 2022 Earnings Conference Call. (Operator Instructions) Please be advised that today's conference is being recorded.

美好的一天，歡迎來到 SoundHound 2022 年第四季度收益電話會議。 （操作員說明）請注意，今天的會議正在錄製中。

I would now like to hand the conference over to your speaker, Mr. Scott Smith, Head of Investor Relations. Please go ahead.

我現在想將會議交給您的發言人，投資者關係主管斯科特史密斯先生。請繼續。

Great. Thank you, Cheri. Hi, everyone. Good afternoon, and thank you for joining our fourth quarter and full year 2022 conference call. With me today is our CEO, Keyvan Mohajer; and our CFO, Nitesh Sharan. We will begin with some short remarks before moving to Q&A.

偉大的。謝謝你，切裡。大家好。下午好，感謝您加入我們的第四季度和 2022 年全年電話會議。今天和我在一起的是我們的首席執行官 Keyvan Mohajer；以及我們的首席財務官 Nitesh Sharan。在進入問答環節之前，我們將從一些簡短的評論開始。

We'd also like to remind everyone that we will be making forward-looking statements on this call. Actual results could differ materially from those suggested by our forward-looking statements. Please refer to our filings with the SEC for a detailed discussion of the risks and uncertainties that could affect our business, and those that qualify as forward-looking statements.

我們還想提醒大家，我們將在本次電話會議上發表前瞻性聲明。實際結果可能與我們的前瞻性陳述所暗示的結果存在重大差異。請參閱我們提交給美國證券交易委員會的文件，詳細討論可能影響我們業務的風險和不確定性，以及那些符合前瞻性陳述條件的風險和不確定性。

In addition, we may discuss certain non-GAAP measures. Please refer to today's press release for more detailed financial results and further details on the definitions, limitations and uses of those measures and reconciliations from GAAP to non-GAAP. Also note that the forward-looking statements on this call are based on information available to us as of today's date. We disclaim any obligation to update any forward-looking statements, except as required by law.

此外，我們可能會討論某些非 GAAP 措施。請參閱今天的新聞稿，了解更詳細的財務結果，以及有關這些措施的定義、限制和使用以及從 GAAP 到非 GAAP 的調節的更多詳細信息。另請注意，本次電話會議的前瞻性陳述是基於截至今天我們可獲得的信息。除非法律要求，否則我們不承擔任何更新任何前瞻性陳述的義務。

Finally, this call is being audio webcast in its entirety on our Investor Relations website. An audio replay will be available shortly following today's call.

最後，我們的投資者關係網站上正在對此次電話會議進行完整的音頻網絡廣播。音頻重播將在今天的電話會議後不久提供。

With that, I'd like to turn the call over to our CEO, Keyvan Mohajer. Please go ahead, Keyvan.

有了這個，我想把電話轉給我們的首席執行官 Keyvan Mohajer。請繼續，Keyvan。

Thank you, Scott, and thank you to everyone for joining the call today. We are happy to share with you the results and highlights of our first fiscal year as a public company, which we closed out with a strong fourth quarter. Our cumulative bookings backlog increased to $332 million at year-end, and our full year 2022 revenue was $31.1 million. Both of these key values exceeded targets we set and communicated in 2021, before we went public and well before the challenging market conditions began.

謝謝斯科特，感謝大家今天加入電話會議。我們很高興與您分享我們作為上市公司的第一個財年的成果和亮點，我們以強勁的第四季度結束了這一財年。到年底，我們的累計預訂積壓增加到 3.32 億美元，我們 2022 年全年的收入為 3110 萬美元。這兩個關鍵價值都超過了我們在 2021 年設定和傳達的目標，那時我們還沒有上市，而且遠在充滿挑戰的市場條件開始之前。

SoundHound's user queries using cloud connectivity were also a highlight, growing over 85% in 2022 compared to '21, resulting in an annual run rate of approximately $2.2 billion. In Q4, our bookings increased 10% quarter-over-quarter, and our revenue was $9.5 million, up 84% year-over-year.

SoundHound 使用雲連接的用戶查詢也是一個亮點，與 21 年相比，2022 年增長了 85% 以上，年運行率約為 22 億美元。第四季度，我們的預訂量環比增長 10%，收入為 950 萬美元，同比增長 84%。

Before diving back in, I want to take a moment to reflect on how we got here and share our view of the latest trends and momentum in AI. We started SoundHound in the Stanford University dorm room, with the vision that conversational AI will be the next major disruption in computing. We knew our AI vision was long term and worked for over a decade to build and own the entire technology stack that is being transformative. We globalized our technology to over 25 languages and are starting the year with $332 million in bookings from our customers. And now we power tens of millions of devices and services around the world, enabling people to more seamlessly interact and exchange in a more accessible interconnected world.

在重新開始之前，我想花點時間回顧一下我們是如何走到這一步的，並分享我們對 AI 最新趨勢和勢頭的看法。我們在斯坦福大學的宿舍裡創辦了 SoundHound，願景是對話式 AI 將成為計算領域的下一個重大顛覆。我們知道我們的 AI 願景是長期的，並且花了十多年的時間來構建和擁有正在變革的整個技術堆棧。我們將我們的技術全球化到超過 25 種語言，並在今年年初從我們的客戶那裡獲得了 3.32 億美元的預訂。現在，我們為全球數以千萬計的設備和服務提供支持，使人們能夠在一個更易於訪問的互聯世界中更無縫地互動和交流。

With the recent popularization of Generative AI and large language models, we view 2023 as an inflection point of conversational AI, with a clear intersection between technology readiness and subsequent demand and adoption. Such intersections are rare and SoundHound is in a unique position to create tremendous value. We believe the impact of Generative AI and large language model to the field of conversational AI is similar to the impact of each neural networks to machine learning. Better quality models can be created faster and with fewer tedious engineering requirements. And we view the market impact of this to be transformative, similar to the mobile ecosystem disruption of the last decade. Fast movers will create tremendous value.

隨著最近生成式 AI 和大型語言模型的普及，我們將 2023 年視為對話式 AI 的轉折點，技術準備與後續需求和採用之間存在明顯的交集。這種交叉點很少見，而 SoundHound 處於創造巨大價值的獨特位置。我們認為生成式人工智能和大型語言模型對對話式人工智能領域的影響類似於每個神經網絡對機器學習的影響。可以更快地創建質量更好的模型，並減少繁瑣的工程要求。我們認為這對市場的影響是變革性的，類似於過去十年的移動生態系統中斷。快速行動者將創造巨大的價值。

Now moving fast does not mean just opening the text or voice interface to a large language model. First models possessed the danger of providing unpredictable, stale and incorrect responses, a phenomenon known as AR hallucinations. The underlying principles of SoundHound technology are addressing these shortcomings by combining software engineering and machine learnings. For the first time, we are unveiling our novel approach using our proprietary technologies called CAILAN and (inaudible).

現在快速發展並不意味著只是打開大型語言模型的文本或語音界面。第一個模型具有提供不可預測的、陳舊的和不正確的反應的危險，這種現像被稱為 AR 幻覺。 SoundHound 技術的基本原則是通過結合軟件工程和機器學習來解決這些缺點。這是我們第一次使用我們稱為 CAILAN 和（聽不清）的專有技術來展示我們的新方法。

CAILAN, or Conversational AI Language uses software engineering to efficiently build knowledge domains, while CAINET or Conversational AI Networks, uses machine learning to augment and enhance the realm of possibilities. CAINET can connect to SoundHound's internal Generative AI models as well as external models, such as OpenAI, GPT and ChatGPT among others. Ultimately, CAILAN controls and arbitrates the results to provide the best response to the user.

CAILAN 或對話式 AI 語言使用軟件工程有效地構建知識領域，而 CAINET 或對話式 AI 網絡則使用機器學習來擴大和增強可能性領域。 CAINET 可以連接到 SoundHound 的內部生成 AI 模型以及外部模型，例如 OpenAI、GPT 和 ChatGPT 等。最終，彩瀾對結果進行控制和仲裁，為用戶提供最佳響應。

Let me demonstrate.

讓我演示一下。

(presentation)

（推介會）

What you just heard was a sneak preview of a new product called SoundHound ChatAI. It ushers in a new phase of bot-enabled conversational AI that used to only exist in science fiction. By combining the power of software engineering and machine learning regenerative AI, we can finally deliver the visual assistant experience that users have been desiring for decades. Please stay tuned for our official launch announcement very soon.

您剛剛聽到的是一款名為 SoundHound ChatAI 的新產品的預覽。它開創了過去只存在於科幻小說中的支持機器人的對話式 AI 的新階段。通過結合軟件工程和機器學習再生人工智能的力量，我們最終可以提供用戶幾十年來一直渴望的視覺助手體驗。請盡快關注我們的官方發佈公告。

Let's review our expected business impact of this change as it relates to our 3 pillar strategy. In our Pillar 1, where we power, devices and cars, more users will use their assistants more often. Users did no longer limit their interactions to simple control command, setting timers and playing music. They can have lengthy, useful conversations with their assistants. This increasing usage will translate to an increase in revenue and licensing fees for SoundHound. Combined with our breakthrough dynamic interaction, which I'll talk about shortly, this will further strengthen our position as a leader in voice interfaces by creating more interaction capabilities for users of our products. In addition, more product creators will demand this feature, and this increasing adoption will also increase the revenue for SoundHound.

讓我們回顧一下這一變化對我們的預期業務影響，因為它與我們的三大支柱戰略相關。在我們提供動力、設備和汽車的支柱 1 中，更多用戶將更頻繁地使用他們的助手。用戶不再將他們的交互僅限於簡單的控制命令、設置定時器和播放音樂。他們可以與助手進行冗長而有益的對話。使用量的增加將轉化為 SoundHound 的收入和許可費用的增加。結合我們突破性的動態交互（我將在稍後討論），這將通過為我們產品的用戶創造更多的交互功能，進一步鞏固我們在語音界面方面的領導地位。此外，更多的產品創造者將需要此功能，而這種日益普及也將增加 SoundHound 的收入。

In our Pillar 2, more tasks can be automated. And in general, customer service, an area of important focus for SoundHound will be transformed with AI being not just cheaper than humans, but also better than humans. By incorporating the possibilities that are already proven from Generative AI and large language models into customer service applications, we envision that end users will ultimately prefer to speak on AI agents, rather than a human representative when talking to a business.

在我們的支柱 2 中，可以自動執行更多任務。總的來說，客戶服務是 SoundHound 的一個重要關注領域，隨著 AI 不僅比人類便宜，而且比人類更好，這一領域將發生轉變。通過將生成式 AI 和大型語言模型已經證明的可能性整合到客戶服務應用程序中，我們預計最終用戶在與企業交談時最終將更喜歡與 AI 代理交談，而不是人類代表。

We expect our expansion to new customer service verticals to accelerate and ultimately, AI-enabled customer service will be as necessary to every business as electricity. SoundHound is in a unique position to take advantage of this rare moment, due to our advanced and comprehensive technologies, mature products and existing customer base. In our view, this can be done as a side project. It can be achieved by simply interpreting a few external APIs and creating the necessary technologies to win in conversational AI would be costly, inefficient and slow. By owning all the core pieces of the engines, we have a distinct competitive advantage.

我們預計我們向新的客戶服務垂直領域的擴張將加速，最終，人工智能支持的客戶服務將像電力一樣成為每個企業所必需的。由於我們先進而全面的技術、成熟的產品和現有的客戶群，SoundHound 處於一個獨特的位置來利用這一難得的時刻。在我們看來，這可以作為一個副項目來完成。它可以通過簡單地解釋一些外部 API 並創建必要的技術來在對話式 AI 中取勝來實現，但這將是昂貴、低效且緩慢的。通過擁有發動機的所有核心部件，我們擁有明顯的競爭優勢。

Innovation has been a core part of our DNA. Last November, we unveiled a new technology called Dynamic Interaction. SoundHound dynamic interaction is a multimodal full duplex interface, with real-time continuous audiovisual feedback, in response to both audio and touch input, with no (inaudible), or turn taking. It ignores off-topic speech, makes for active suggestions to the users and intelligently decides when to use audio or visual output.

創新一直是我們 DNA 的核心部分。去年 11 月，我們推出了一項名為動態交互的新技術。 SoundHound 動態交互是一種多模式全雙工界面，具有實時連續的視聽反饋，響應音頻和触摸輸入，沒有（聽不清）或輪流。它忽略離題的演講，為用戶提供積極的建議，並智能地決定何時使用音頻或視頻輸出。

We believe dynamic interaction is a category-defining breakthrough that will raise the bar for human computer interaction, just like how Apple's multitouch technology (inaudible) existing touch interfaces in 2007. By combining dynamic interaction with Generative AI, we expect to achieve our ultimate vision of making computers better than humans in language understanding and more human in the way they interact and respond.

我們相信動態交互是一個定義類別的突破，它將提高人機交互的標準，就像 Apple 的多點觸控技術（聽不清）在 2007 年現有的觸摸界面一樣。通過將動態交互與生成 AI 相結合，我們期望實現我們的最終願景讓計算機在語言理解方面比人類更好，在交互和響應方面更人性化。

Given our unique positioning and strength, the growing demand we are seeing for our independent AI powered products and services will only get stronger, as we execute on more opportunities and with more customers.

鑑於我們獨特的定位和實力，我們看到對我們獨立的 AI 驅動產品和服務的需求不斷增長，因為我們會抓住更多機會並與更多客戶合作。

In 2022, we made great progress in forging and extending significant relationships and partnerships with a number of new brands with our vertical agnostic technology. In the past year, SoundHound has announced a number of key partnerships and integrations, including an agreement with Qualcomm to bring SoundHound's voice AI to its Snapdragon platforms, new deals and integrations with VIZIO, Square, Toast and Oracle, and an expanded partnership with Snap. SoundHound also signed a multiyear agreement with Hyundai, and announced an expansion of our Stellantis relationship in Europe, as well as a series of collaboration in automotive space, including with LG, HARMAN, DPCA and DMI.

2022 年，我們利用垂直不可知論技術在與許多新品牌建立和擴展重要關係和合作夥伴關係方面取得了長足進步。在過去的一年裡，SoundHound 宣布了多項重要的合作夥伴關係和整合，包括與高通達成協議，將 SoundHound 的語音 AI 引入其驍龍平台，與 VIZIO、Square、Toast 和 Oracle 的新交易和整合，以及與 Snap 的擴大合作夥伴關係. SoundHound 還與現代汽車簽署了一項多年協議，並宣布擴大我們在歐洲的 Stellantis 關係，以及在汽車領域的一系列合作，包括與 LG、HARMAN、DPCA 和 DMI 的合作。

In automotive, we now work with 20 key brands, an impressive number given we went from 0 to 20 brands in just a few years. While we continue to add new brands in our portfolio, we are also able to expand with our existing brands globally, by supporting new languages, new regions, adding edge and cloud and new services. We are also working with many world-class device manufacturers ranging from smart appliances to TVs and in multiple areas within the IoT space. Here, we continue to expand within our existing customers, and we are excited about the companies in our pipeline. With more than 75 billion connected devices expected globally by 2025, we see a tremendous opportunity here for our most AI powered products and solutions.

在汽車領域，我們現在與 20 個主要品牌合作，考慮到我們在短短幾年內從 0 個品牌發展到 20 個，這個數字令人印象深刻。在我們繼續在我們的產品組合中添加新品牌的同時，我們還能夠通過支持新語言、新地區、添加邊緣和雲以及新服務，在全球範圍內擴展我們現有的品牌。我們還與許多世界級的設備製造商合作，從智能家電到電視，以及物聯網領域的多個領域。在這裡，我們繼續在現有客戶中擴展，我們對我們管道中的公司感到興奮。到 2025 年，全球預計將有超過 750 億台聯網設備，我們在這裡看到了我們最受人工智能支持的產品和解決方案的巨大機遇。

Last year, our voice AI technology was made generally available for customer service, starting with SoundHound for Restaurants, making our entry into this massive rapidly evolving service industry. Since launch, we have added restaurant customers across North America and continue to forge relationships with important partners, including companies like Oracle, Square and Toast, with more to come.

去年，我們的語音 AI 技術普遍可用於客戶服務，從 SoundHound for Restaurants 開始，使我們進入了這個快速發展的龐大服務行業。自推出以來，我們在北美增加了餐廳客戶，並繼續與重要合作夥伴建立關係，包括 Oracle、Square 和 Toast 等公司，未來還會有更多合作夥伴。

Our SoundHound for restaurant voice AI Assistant allows restaurants to automate their ordering process at drive-throughs, ordering kiosks, in-app and over the phone, enabling restaurants to create efficiencies in order taking, at a time when many are tackling rising food costs, staffing shortages, supply chain issues and other headwinds. Voice-enabled food order is particularly challenging, requiring specialized speech recognition and complex natural language understanding, which matches perfectly with our key differentiators.

我們用於餐廳語音 AI 助手的 SoundHound 允許餐廳在免下車點餐、點餐亭、應用程序內和電話上實現點餐流程自動化，使餐廳能夠在許多人應對不斷上漲的食品成本的情況下提高點餐效率，人員短缺、供應鏈問題和其他不利因素。支持語音的食品訂單特別具有挑戰性，需要專門的語音識別和復雜的自然語言理解，這與我們的關鍵差異化因素完美匹配。

We are excited as we move forward in 2023, because we expect to continue to add customers and partners and deliver voice AI technology to some of the most well-known brands in the world. We aim to power billions of devices and services, and with already voice-enabled millions of cars, TVs, mobile apps, IOT devices and restaurants on a global scale, thanks to our expansive set of languages.

我們很高興在 2023 年繼續前進，因為我們希望繼續增加客戶和合作夥伴，並為世界上一些最知名的品牌提供語音 AI 技術。我們的目標是為數十億台設備和服務提供支持，並且已經在全球範圍內為數百萬輛支持語音的汽車、電視、移動應用程序、物聯網設備和餐廳提供支持，這要歸功於我們廣泛的語言集。

And while we are excited about what we have built at SoundHound and the new opportunities we have realized, it is nice to see that this asset is also being recognized by the industry. Recently, we have been named as Speech Industry Award Winner, and we were also named as one of the world's top 10 disruptive AI companies by AI magazine. In addition, SoundHound for Restaurants has been named among the winners of the National Restaurant Association, Smartbrief Innovation Award for foodservice, after only a short period since market launch.

雖然我們對 SoundHound 的成就和我們實現的新機遇感到興奮，但很高興看到該資產也得到了業界的認可。最近，我們被評為 Speech Industry Award Winner，我們也被 AI 雜誌評為全球十大顛覆性 AI 公司之一。此外，SoundHound for Restaurants 在推出市場後不久就被評為全國餐飲協會 Smartbrief 餐飲服務創新獎的獲獎者之一。

There is undoubtedly momentum building behind AI, our company and our category. The convergence of heightened market demand for conversational AI and the technological readiness for our AI powered solutions has positioned us at a critical and a very exciting inflection point, one we are primed to take advantage of.

毫無疑問，在人工智能、我們公司和我們的品類背後，勢頭正在增強。市場對對話式 AI 需求的增加與我們 AI 驅動的解決方案的技術準備情況的融合使我們處於一個關鍵且非常令人興奮的轉折點，我們已準備好利用這一轉折點。

In closing, since launching SoundHound's initial voice AI platform in 2015, we have radically evolved it, globalized its reach to 25 languages, been selected by 20 car brands and created over 100 branded assistants across numerous brands and languages. We continue to scale into major enterprises, including Hyundai, Kia, Genesis, Mercedes-Benz, Stellantis, Dodge, Chrysler, HARMAN, Toast, Square, LG, VIZIO, Snap, Pandora and many others.

最後，自 2015 年推出 SoundHound 的初始語音 AI 平台以來，我們對其進行了徹底的改進，將其覆蓋範圍擴展到 25 種語言，被 20 個汽車品牌選中，並創建了 100 多個品牌和語言的品牌助手。我們繼續擴展到主要企業，包括現代、起亞、Genesis、梅賽德斯-奔馳、Stellantis、Dodge、克萊斯勒、HARMAN、Toast、Square、LG、VIZIO、Snap、Pandora 等。

There is tremendous opportunity in itself to expand within these existing customers, but we are also making meaningful progress in expanding our addressable market, as demonstrated in SoundHound for Restaurants in customer service. And with recent product launches such as dynamic interaction for customer service and Generative AI, we are confident in our ability to both grow within our existing global addressable market and expand into new ones.

在這些現有客戶中進行擴展本身就有巨大的機會，但我們也在擴大我們的目標市場方面取得了有意義的進展，正如 SoundHound for Restaurants 在客戶服務中所展示的那樣。隨著最近產品的推出，例如客戶服務的動態交互和生成人工智能，我們有信心在現有的全球可尋址市場中發展並擴展到新市場。

We are in a unique position to maintain our leadership in voice-enabled AI technology and solve extremely difficult problems with our breakthrough inventions for our customers and partners. We are pleased to have ended the year at the high end of our guidance, despite the macroeconomic conditions that worsened throughout the year. We remain agile and focused and are confident to grow the business at approximately 50% in 2023.

我們處於獨特的地位，可以保持我們在支持語音的 AI 技術方面的領先地位，並通過我們的突破性發明為我們的客戶和合作夥伴解決極其困難的問題。儘管全年宏觀經濟狀況惡化，但我們很高興在我們指導的高端結束了這一年。我們保持敏捷和專注，並有信心在 2023 年將業務增長約 50%。

With that, I will now turn the call over to Nitesh, to talk about our financial performance for the quarter.

有了這個，我現在將把電話轉給 Nitesh，談談我們本季度的財務業績。

Thank you, Keyvan, and good afternoon, everyone. 2022 was our first fiscal year in the public market, a year when the IPO markets stalled, higher interest rates steered investors towards value, and sentiment shifted away from disruptive, high-growth technology companies such as ourselves. Nevertheless, we adjusted, fought harder aggressively took actions and set ourselves up to serve the rapidly increasing demand for our products, and on much stronger financial footing.

謝謝你，Keyvan，大家下午好。 2022 年是我們在公開市場上的第一個財政年度，這一年 IPO 市場停滯不前，更高的利率引導投資者轉向價值，人們的情緒從像我們這樣的顛覆性、高增長的科技公司轉移。儘管如此，我們進行了調整，更加積極地採取了行動，並為滿足對我們產品快速增長的需求而做好了準備，並建立了更強大的財務基礎。

Artificial intelligence is fundamentally about making technology more human-like. At SoundHound, we do that by making conversations with machines more natural and seamless, and the demand for that has never been stronger. Large language models, Generative AI and the pervasiveness of chatbot are unearthing entirely new used cases every day. The opportunity set and adoption curves will accelerate rapidly. As Keyvan indicated, our time is now.

人工智能從根本上說是讓技術更像人類。在 SoundHound，我們通過使與機器的對話更加自然和無縫來做到這一點，對此的需求從未如此強烈。大型語言模型、生成式 AI 和聊天機器人的普及每天都在挖掘全新的用例。機會集和採用曲線將迅速加速。正如 Keyvan 所指出的，我們的時間到了。

I'll dive deeper into expectations for 2023 shortly, let me first review our 2022 results. We continue to expand our voice-enabled ecosystem, have strong growth and delivered full year revenue at the high end of our guidance range. Despite the volatile macro backdrop, we have now outperformed our publicly stated full year outlooks for the past 2 years. The foundation for this consistent overachievement versus our expectations is rooted in strong customer engagement and sound execution, that should only further improve from here.

我將很快深入探討對 2023 年的預期，讓我首先回顧一下我們 2022 年的結果。我們繼續擴大我們的語音生態系統，實現強勁增長，並在我們指導範圍的高端實現全年收入。儘管宏觀背景動盪不安，但我們過去 2 年的表現優於我們公開聲明的全年展望。與我們的預期相比，這種始終如一的超額成就的基礎植根於強大的客戶參與度和良好的執行力，這應該只會從這裡進一步改進。

Our cumulative bookings backlog establishes that foundation. We closed the year with $332 million, representing a year-over-year growth of 59%. The contracts underlying our bookings backlog now range from 1 year to more than 9 years, with a roughly 6.5 year weighted average contract length. Through the year, we saw meaningful lengthening in customer contracts, which demonstrates the deepening strategic partnerships and validates the value we continue to create.

我們累積的預訂積壓建立了這個基礎。我們以 3.32 億美元結束了這一年，同比增長 59%。我們的預訂積壓合同現在從 1 年到 9 年以上不等，加權平均合同期限約為 6.5 年。在這一年裡，我們看到客戶合同顯著延長，這表明戰略合作夥伴關係不斷加深，並驗證了我們繼續創造的價值。

Our business model is grounded in the 3-pillar revenue framework; pillar 1 represents voice-enabled products, where we receive royalties. Pillar 2 represents voice-enabled services, generally under monthly subscription contracts. And then we bring Pillar 1 and 2 together into Pillar 3, monetization.

我們的商業模式基於三支柱收入框架；支柱 1 代表語音產品，我們從中收取版稅。支柱 2 代表語音服務，通常按月訂閱合同。然後我們將支柱 1 和支柱 2 合併為支柱 3，貨幣化。

As we have discussed previously, the bulk of our current business and backlog is in pillar 1 and in 2022, we scaled into millions of new cars and devices. In this pillar, we are extending our offering across new units and adding more features to existing ones to expand revenue per unit. We can scale with existing customers, and we are constantly adding new ones.

正如我們之前所討論的，我們當前的大部分業務和積壓訂單都在支柱 1 中，到 2022 年，我們將擴展到數百萬輛新汽車和設備。在這一支柱中，我們正在將我們的產品擴展到新的單元，並為現有單元添加更多功能，以擴大每單元的收入。我們可以擴展現有客戶，並且不斷增加新客戶。

In automotive, for example, we saw more than 3x growth in, in year new unit production in 2022 and unit price expansion of more than 20%. And in Q4, we added another 3 brands to our ever-growing list.

例如，在汽車行業，我們看到 2022 年新單位產量增長了 3 倍以上，單價漲幅超過 20%。在第四季度，我們將另外 3 個品牌添加到我們不斷增長的列表中。

As we have discussed previously, growing our Pillar 2 voice-enabled services is a key focus area, starting with AI-enabled customer services for restaurant. There's massive demand and one advantage here, is the pace at which we can scale, because of the much shorter sales cycle and activation timeline. Just several months in and our qualified advanced pipeline of stores is now well into the thousands, and we are rapidly enhancing product features to extend our reach to even more brands and locations.

正如我們之前所討論的，發展我們的支柱 2 語音服務是一個關鍵的重點領域，首先是餐廳的 AI 客戶服務。這裡有大量的需求和一個優勢，就是我們可以擴展的速度，因為銷售週期和激活時間表要短得多。僅僅幾個月後，我們合格的高級商店管道現已達到數千家，並且我們正在迅速增強產品功能，以將我們的影響力擴展到更多品牌和地點。

The recent advances with large language models only accelerates our excitement around the broader customer service opportunity in this pillar. Our packaged offerings and technologically advanced business inquiry voice AI solutions, are perfectly suited to extend our reach to a multitude of businesses and industries. There is tremendous value to unlock for customers.

大型語言模型的最新進展只會加速我們對該支柱中更廣泛的客戶服務機會的興奮。我們的打包產品和技術先進的業務查詢語音 AI 解決方案非常適合將我們的業務範圍擴展到眾多企業和行業。為客戶解鎖巨大的價值。

Let me now get specific on our financial results for the fourth quarter and full year. In Q4, we generated $9.5 million in revenue, up 84% year-over-year. For the full year, our revenue was $31.1 million, and was predominantly driven by product royalties, with an increasing contribution from subscription services throughout the year. Our product royalty revenue increased primarily due to strong customer momentum and a number of large edge deals, where we received multiyear minimum guarantee commitments.

現在讓我具體談談我們第四季度和全年的財務業績。第四季度，我們創造了 950 萬美元的收入，同比增長 84%。全年，我們的收入為 3110 萬美元，主要由產品版稅推動，全年訂閱服務的貢獻不斷增加。我們的產品特許權使用費收入增加主要是由於強勁的客戶勢頭和一些大型邊緣交易，我們在這些交易中獲得了多年最低保證承諾。

Our edge solutions operate without Internet connectivity or hosting requirements. So our performance obligations are fulfilled when we deliver the software. Our long-term commitments demonstrate the continued strong partnerships we've developed and the continued share gains we are experiencing. Foreign exchange also had an impact on our revenue. Our full year revenue growth rate would have been roughly 2 percentage points higher without this impact.

我們的邊緣解決方案無需互聯網連接或託管要求即可運行。因此，當我們交付軟件時，我們的履約義務就得到了履行。我們的長期承諾證明了我們已經建立的持續牢固的合作夥伴關係以及我們正在經歷的持續份額增長。外匯也對我們的收入產生了影響。如果沒有這種影響，我們的全年收入增長率將高出大約 2 個百分點。

In Q4, our gross margin improved to 71%, up from 67% in the prior year quarter, which was largely driven by the expanding scale of our business. Cost of revenue for the quarter was $2.8 million, up 61% from the prior year. The majority of our cost of revenue includes data center costs supporting our customer production environment.

第四季度，我們的毛利率從去年同期的 67% 提高到 71%，這主要是由於我們業務規模的擴大。本季度的收入成本為 280 萬美元，比去年同期增長 61%。我們的大部分收入成本包括支持我們客戶生產環境的數據中心成本。

We completed the migration of our cloud [spin] to Oracle in Q4 2022. As we look forward, there are some on-prem activities that we (inaudible) related to the cloud, helping us continue to drive gross margin expansion along with the anticipated benefits of further scale.

我們在 2022 年第四季度完成了雲 [spin] 到 Oracle 的遷移。展望未來，我們（聽不清）有一些與雲相關的本地活動，幫助我們繼續推動毛利率擴張以及預期進一步擴大規模的好處。

Moving on to operating expenses. R&D has been the key pillar of this company, as we built our voice AI platform and deep patent portfolio of over 260 filed and pending. In Q4, R&D was $21.5 million, up 32% year-over-year. We continue to invest in R&D to ensure we remain at the forefront of innovation in AI and machine learning, while also helping to develop and scale our cloud offerings in other products and services.

繼續經營費用。研發一直是這家公司的關鍵支柱，因為我們建立了語音人工智能平台和超過 260 項已申請和正在申請的深度專利組合。第四季度，研發費用為 2150 萬美元，同比增長 32%。我們繼續投資於研發，以確保我們始終處於人工智能和機器學習創新的前沿，同時也幫助開發和擴展我們在其他產品和服務中的雲產品。

In 2022, we ramped up our sales and marketing investments in both people and technology to accelerate growth, particularly in pillar 2. We have increased focus in digital marketing, lead generation and customer acquisitions, both direct and through channel partners. In Q4, sales and marketing expenses were $6.7 million, up 587% year-over-year, off a small prior year base. With our organizational shifts in 2023, we expect this expense item to benefit from greater focus.

2022 年，我們加大了對人員和技術的銷售和營銷投資，以加速增長，尤其是在支柱 2 中。我們更加關注數字營銷、潛在客戶開發和客戶獲取，包括直接和通過渠道合作夥伴。第四季度，銷售和營銷費用為 670 萬美元，同比增長 587%，低於去年同期的小幅基數。隨著我們在 2023 年的組織結構轉變，我們預計這一支出項目將受益於更多的關注。

G&A expenses were $7.2 million in Q4, up 41% year-over-year, which reflects investments across the global support functions of finance, legal, facilities and human resources. The increases were largely a result of ensuring our G&A functions delivered on the requirements of a newly public company.

第四季度的 G&A 費用為 720 萬美元，同比增長 41%，這反映了對財務、法律、設施和人力資源等全球支持職能的投資。增加的主要原因是確保我們的 G&A 職能能夠滿足新上市公司的要求。

Across all operating expenses, noncash employee stock compensation was $9.3 million in Q4. We expect the ongoing quarterly stock comp expense to meaningfully reduce in 2023, post our recent restructuring. Our operating loss of $28.8 million in Q4 and adjusted EBITDA, which excludes the noncash charges of stock compensation and depreciation and amortization and other nonoperating activities, was a loss of $18.6 million.

在所有運營費用中，第四季度的非現金員工股票薪酬為 930 萬美元。在我們最近的重組之後，我們預計持續的季度股票補償費用將在 2023 年顯著減少。我們在第四季度的經營虧損為 2880 萬美元，調整後的 EBITDA（不包括股票補償、折舊和攤銷以及其他非經營活動的非現金費用）為 1860 萬美元。

Net loss per share in Q4 was $0.15 compared to a net loss of $0.32 in the prior year period. For the full year, net loss per share was $0.83 compared to a net loss of $1.18 in the prior year.

第四季度每股淨虧損為 0.15 美元，而去年同期為每股淨虧損 0.32 美元。全年每股淨虧損為 0.83 美元，而去年同期為 1.18 美元。

On to the capital structure; our cash position at year-end was $9.2 million. In January, we successfully raised $25 million in preferred equity. Our previously announced committed equity line of credit also became effective, giving us additional access to capital. In addition, as we announced concurrently with our preferred rates, we are in process of refinancing our existing debt. All in, our balance sheet has strengthened meaningfully and we have a fully funded business plan through and beyond the point of our becoming profitable.

關於資本結構；年末我們的現金頭寸為 920 萬美元。 1 月，我們成功籌集了 2500 萬美元的優先股。我們先前宣布的承諾股權信貸額度也已生效，使我們能夠獲得更多資金。此外，正如我們與首選利率同時宣布的那樣，我們正在為現有債務再融資。總而言之，我們的資產負債表得到了顯著增強，並且我們有一個資金充足的商業計劃，可以實現盈利。

With that, I'll move on to guidance. Earlier this year, we announced restructuring actions that were necessary to right-size our cost structure, impacting both people and discretionary spend. Such decisions are never easy, but the actions have enabled us to accelerate our path to profitability without materially impacting our ability to execute our business plan effectively. We expect these cost reductions to result in operating cost savings of approximately $60 million in 2023, as compared to 2022. We expect revenue to grow approximately 50% in 2023, and we now expect to be adjusted EBITDA positive in Q4 of this year, setting the stage for sustained long-term profitable growth, as we enter 2024.

有了這個，我將繼續指導。今年早些時候，我們宣布了調整成本結構所必需的重組行動，這對人員和可自由支配的支出都有影響。這樣的決定絕非易事，但這些行動使我們能夠在不對我們有效執行業務計劃的能力產生實質性影響的情況下加快盈利之路。與 2022 年相比，我們預計這些成本削減將在 2023 年節省約 6000 萬美元的運營成本。我們預計 2023 年收入將增長約 50%，我們現在預計今年第四季度調整後的 EBITDA 為正，設定隨著我們進入 2024 年，持續的長期盈利增長階段。

Let me double-click a bit on these expectations. First, the $332 million of cumulative bookings backlog that I noted earlier, gives us high confidence on revenue that will either automatically roll into the P&L in accordance with software revenue recognition guidance, or as a result of expected customer unit volumes, upon which we generate royalty revenue streams. In addition, we expect that SoundHound for Restaurants will provide a meaningful contribution to revenue and grow at an accelerated rate, as we move through the year. It is important to note that SoundHound for Restaurants has had only a minimal contribution to our revenue up to this point, but the underlying customer foundation pipeline and activation of stores is building rapidly. It's also important to note that this vertical has a very attractive financial profile, with more predictable revenue streams, better scalability, leading to more favorable operating margins and a fast conversion of initial customer contact to cash flow.

讓我雙擊這些期望。首先，我之前提到的 3.32 億美元的累積預訂積壓讓我們對收入充滿信心，這些收入將根據軟件收入確認指南自動轉入損益表，或者作為我們生成的預期客戶單位量的結果版稅收入流。此外，我們預計 SoundHound for Restaurants 將為收入做出有意義的貢獻，並在我們度過這一年時以更快的速度增長。值得注意的是，到目前為止，SoundHound for Restaurants 對我們的收入貢獻微乎其微，但潛在的客戶基礎管道和商店的激活正在迅速建立。同樣重要的是要注意，這個垂直行業具有非常有吸引力的財務狀況，具有更可預測的收入流、更好的可擴展性，從而帶來更有利的營業利潤率以及從初始客戶聯繫到現金流的快速轉換。

We are enthusiastic about the momentum we continue to see from customers and how our team is capitalizing on this tremendous opportunity. Therefore, overall in 2023, we expect that revenue will be in the range of 43 and $50 million. We expect this revenue will build through the year, with each quarter larger than the last, due to seasonality of Pillar 1 businesses and the scaling of restaurants. Our Q1 results will also be impacted by the restructuring efforts that we continue to finalize. Last year, approximately 2/3 of our revenue was recognized in the second half of the year, and we expect to see a similar trend for this year.

我們對我們繼續從客戶那裡看到的勢頭以及我們的團隊如何利用這一巨大機會充滿熱情。因此，總體而言，2023 年，我們預計收入將在 43 至 5000 萬美元之間。由於支柱 1 業務的季節性和餐廳的規模擴大，我們預計這一收入將在全年增加，每個季度都比上一個季度增加。我們的第一季度業績也將受到我們繼續完成的重組工作的影響。去年，我們大約 2/3 的收入是在下半年確認的，我們預計今年也會出現類似的趨勢。

To summarize, we have taken the important steps to set us up for long-term success. While our progress will not always be linear, we are building a strong and sustainable business model, and driving growth with a clear path to profitability. Market momentum is real, customer traction is strong, and we are extremely excited about our path forward.

總而言之，我們已經採取了重要的步驟來為我們的長期成功做好準備。雖然我們的進展不會總是線性的，但我們正在建立一個強大且可持續的商業模式，並通過明確的盈利途徑推動增長。市場勢頭真實，客戶吸引力強勁，我們對前進的道路感到非常興奮。

Thank you. And we will now move to Q&A.

謝謝。我們現在將進入問答環節。

(Operator Instructions) And today's first question will come from the line of Mike Latimore with Northland.

（操作員說明）今天的第一個問題將來自 Northland 的 Mike Latimore。

Congrats on the excellent results this year. So a lot of interesting developments, obviously. I guess, first on the restaurant vertical, you seem fairly positive about that vertical. I guess, can you just give a little more color there? I mean, how many restaurants are you currently deployed at, and when you talk about pipeline, is that a business that's about to be deployed, or is that a business that you still need to win? And then I guess also, it just seems like this would be applicable to a very large chains, are there any large restaurant chains that are rolling this out?

祝賀今年的優異成績。顯然，有很多有趣的發展。我想，首先在餐廳垂直領域，你似乎對該垂直領域相當樂觀。我想，你能給那裡多一點顏色嗎？我的意思是，您目前部署了多少家餐廳，當您談論管道時，這是一項即將部署的業務，還是您仍需要贏得的業務？然後我想，這似乎適用於非常大的連鎖店，是否有任何大型連鎖餐廳正在推出這個？

Sure, Mike. I could start. So a couple of points. Yes, when I alluded to the pipeline in my prepared remarks, I mean, generally, we -- obviously, we have a lot of conversations going on, and we sort of have an initial pipeline and then we kind of enter it into our CRS systems when it's in a more advanced stage. So qualified leads that we really characterize on what I was referencing, are ones where there are deep conversations going on with decisionmakers at the restaurant, where there's clear line of sight to a significant opportunity, whether it's small and medium enterprises, it could be 1 to a single-digit level location. In the enterprise side, they could be tens, hundreds plus, and we definitely have a suite of both small, medium, mid-market and enterprise-level customers that have hundreds of stores.

當然，邁克。我可以開始了。所以有幾點。是的，當我在準備好的發言中提到管道時，我的意思是，一般來說，我們 - 顯然，我們正在進行很多對話，我們有一個初步的管道，然後我們將它輸入我們的 CRS處於更高級階段的系統。如此合格的潛在客戶，我們真正根據我所引用的內容進行了描述，這些潛在客戶與餐廳的決策者進行了深入對話，對重要機會有清晰的視線，無論是中小型企業，都可能是 1到個位數級別的位置。在企業方面，他們可能有數十家、數百家，我們肯定擁有一套擁有數百家商店的小型、中型、中型市場和企業級客戶。

We've previously mentioned White Castle as an example, a chain that has over 350 stores around the nation, and we're activating it live in several of those and scaling. I would say that in terms of the number, I have mentioned thousands of advanced stage pipeline. So those are ones that are not yet paying us monthly fee, those are pipeline. So that means they're not yet in, but those are late-stage conversations.

我們之前提到過 White Castle 作為一個例子，這是一家在全國擁有超過 350 家商店的連鎖店，我們正在其中幾家商店中激活它並擴大規模。我要說的是，就數量而言，我已經提到了數千個高級階段流水線。因此，那些尚未向我們支付月費的是那些正在籌備中的。所以這意味著他們還沒有加入，但那些是後期對話。

In many cases, what happens is, we kind of get a late-stage conversation with, I'd say, a pizza restaurant and there are certain features or facets that they need, maybe they need a customization of doing half this half that, and that's a feature that we developed. So in a lot of cases, there's really active engagement, we may just need to add some features and functionality.

在許多情況下，發生的情況是，我們與一家比薩餅餐廳進行了後期對話，他們需要某些功能或方面，也許他們需要定制化來做一半，這是我們開發的一項功能。所以在很多情況下，確實有積極的參與，我們可能只需要添加一些特性和功能。

We have hundreds of restaurants signed up already. So beyond the pipeline stage that are operating, and we're quickly scaling those. So as I mentioned in the prepared remarks, the time from sort of initial conversation to cash flow or monthly recurring revenue indicates for us, can be very rapid. And especially when you get into restaurants that have multiple stores, because once you ingest that menu upfront and you custom sort of develop the capability and the voice, moving to restaurant #2 or #3, is very, very fast. In some cases, if it's more on the drive-through feature, you might have some acoustic things that you have to figure out, but generally, especially for cornering, you don't have any of those complexities of traffic or outside noise, and so it could be much more rapid.

我們已經有數百家餐廳簽約。因此，在正在運行的管道階段之外，我們正在快速擴展這些階段。因此，正如我在準備好的評論中提到的那樣，從最初的談話到現金流或每月經常性收入對我們來說表明的時間可能非常快。尤其是當您進入擁有多家商店的餐廳時，因為一旦您預先攝取了該菜單並且您自定義了某種開發能力和聲音，移動到 #2 或 #3 餐廳，速度非常非常快。在某些情況下，如果更多的是關於直通車功能，您可能需要解決一些聲學問題，但通常，尤其是在轉彎時，您不會遇到交通或外部噪音的任何復雜情況，並且所以它可能會更快。

I rambled a little bit there, so maybe I'll pause and see if that kind of hits the question.

我在那兒漫無目的，所以也許我會停下來看看是否能解決問題。

Yes. That's good context. And then on the Chat AI service you mentioned, is that both a direct-to-consumer and business enabler type of service, or how are you thinking about kind of the go-to-market there?

是的。這是很好的背景。然後關於您提到的 Chat AI 服務，它是直接面向消費者和業務推動者的服務類型，還是您如何考慮進入市場的類型？

Yes. There will be both, to offer that to our existing customers like car makers and IT makers and so on. And SoundHound, historically, we've had direct-to-consumer applications and products, and our existing applications have received over 300 million downloads. So that's something that is part of our DNA, we're really good at it, and it's a good moment to take advantage of that experience.

是的。兩者都會提供給我們現有的客戶，如汽車製造商和 IT 製造商等。而 SoundHound，從歷史上看，我們擁有直接面向消費者的應用程序和產品，我們現有的應用程序已獲得超過 3 億次下載。所以這是我們 DNA 的一部分，我們真的很擅長，現在是利用這種經驗的好時機。

Great. And just last one, can you provide a little more color on your activities with Qualcomm, that seemed like an important partner that you announced last year, maybe just a little bit more feel for how that partnership is going?

偉大的。最後一個，你能否提供更多關於你與高通的活動的顏色，這似乎是你去年宣布的一個重要合作夥伴，也許只是對這種夥伴關係的進展有更多的感覺？

Yes, so we had a really good demonstration. We had a joint demo at their Snapdragon Summit a few months ago. So -- and they also showcased it at CES. So they are basically bundling our Voice AI solution into their chipsets and they can basically be our channel partner and sell it to the customers on our behalf. So it's a combination of their chipsets and their voice technologies, voice [recognition] technologies and our voice (inaudible) solutions. So it's a very meaningful partnership.

是的，所以我們有一個非常好的演示。幾個月前，我們在他們的 Snapdragon 峰會上進行了聯合演示。所以 - 他們還在 CES 上展示了它。所以他們基本上是將我們的語音人工智能解決方案捆綁到他們的芯片組中，他們基本上可以成為我們的渠道合作夥伴並代表我們將其銷售給客戶。所以這是他們的芯片組和他們的語音技術、語音 [識別] 技術和我們的語音（聽不清）解決方案的結合。所以這是一個非常有意義的伙伴關係。

(Operator Instructions) One moment for our next question, and that will come from the line of Brett Knoblauch with Cantor Fitzgerald.

（操作員說明）請稍等一下我們的下一個問題，這將來自 Brett Knoblauch 和 Cantor Fitzgerald 的台詞。

The first, I guess, is on the profitability guide. I know you guys are taking a lot of actions to kind of improve profitability over the last couple of months. What assumptions do you have to get to kind of positive adjusted EBITDA in the fourth quarter? I guess, what should we be expecting from a gross margin front, absolute OpEx basis, if we look at the run rate exiting the quarter, should we expect that to decline by, call it, $10 million on a run rate basis, or what should we be thinking for that?

我想，首先是關於盈利能力指南。我知道你們在過去幾個月採取了很多行動來提高盈利能力。您對第四季度調整後的 EBITDA 有何假設？我想，如果我們看一下本季度結束的運行率，我們應該從毛利率、絕對運營支出的基礎上期待什麼，我們是否應該期望它下降 1000 萬美元（按運行率計算），或者什麼？我們應該為此考慮嗎？

Sure, Brett. So first, I'll start at the top. So we guided 50% growth and sort of highlighted that, that is going to be back-end loaded. So we are expecting to ramp through the year. Part of that is, just seasonality that comes with the Pillar 1 and predominantly auto business, but number 2 -- well, I should say, auto and devices. So there is seasonality associated with like TV sales and so forth as well. But then also with our Pillar 2 in restaurants as we continue to scale, gain traction, and that builds on a monthly subscription basis, that will compound in a positive way that should build through the year. So there's a bit of that seasonality.

當然，布雷特。所以首先，我將從頭開始。因此，我們指導了 50% 的增長，並強調了後端加載。因此，我們預計今年會有所增長。部分原因是第一支柱和主要汽車業務帶來的季節性，但第二支柱——嗯，我應該說，汽車和設備。因此，電視銷售等也存在季節性。但隨著我們繼續擴大規模，獲得牽引力，並且建立在每月訂閱的基礎上，我們在餐廳的支柱 2 也將以積極的方式複合，這將在今年建立起來。所以有一點季節性。

And then with margins, we're not really guiding margins, but I'd say that just overall profile and long term, we definitely think the levels, the 70% plus level that we've been delivering this year is very achievable and very scalable. As you go into our diversification of products from cloud to edge that we've seen this year, that does have favorable elements to gross margin. I mentioned in my prepared remarks, through the year, we were migrating cloud, and we had that sort of impact, a little bit of a bit of a drag in 2022's margin, that is now kind of completed. And then also there's some other actions we're taking to drive efficiencies in margin and then also just with scale. But with respect to cloud versus Edge, some of the Edge solutions, they tend to have favorable gross margin profile. So as you put it all together, there's -- we expect gross margin to continue to be in the zone and positive to what it was in 2022.

然後關於利潤率，我們並不是真正的指導利潤率，但我想說的是總體概況和長期來看，我們肯定認為我們今年提供的水平，即 70% 以上的水平是非常可以實現的，而且非常可擴展的。當您進入我們今年看到的從雲到邊緣的產品多元化時，這確實對毛利率有利。我在準備好的發言中提到，這一年來，我們一直在遷移雲，我們產生了那種影響，有點拖累了 2022 年的利潤率，現在已經完成了。然後還有一些我們正在採取的其他行動來提高利潤效率，然後也只是擴大規模。但相對於雲與邊緣，一些邊緣解決方案，它們往往具有良好的毛利率。因此，當你把所有這些放在一起時，我們預計毛利率將繼續保持在該區域，並與 2022 年的水平持平。

And then on the OpEx line, I mentioned on a full year basis, $60 million of savings. And the first quarter will be -- we'll have some of the restructuring costs impacting it. But on an ongoing -- and we're still -- for some of our international countries, it just takes time to kind of work through that. But certainly, by the end of the year, we'll have worked through entirely. In fact, in Q1 by the time we're done with Q1, we should be largely through all the restructuring activities. So we should start to realize those savings in OpEx.

然後在 OpEx 線上，我提到全年節省 6000 萬美元。第一季度將是——我們將有一些重組成本影響它。但是，對於我們的一些國際國家來說，正在進行的——我們仍然是——需要時間來解決這個問題。但可以肯定的是，到今年年底，我們將完全完成工作。事實上，當我們完成第一季度時，我們應該在第一季度完成所有重組活動。因此，我們應該開始在運營支出中實現這些節省。

And yes, I mean, as you get to the full year Q4 estimation, I think when you look at our OpEx items, R&D, sales and marketing, G&A, where we're taking and the actions we took were north of 40%-plus reductions on -- both head count and discretionary, you can kind of do the math. The other thing I'd say -- and I mentioned this in the prepared remarks impacting as well around the stock comp elements, although that doesn't affect adjusted EBITDA, but that also, we expect to come down.

是的，我的意思是，當你看到第四季度的全年估計時，我認為當你查看我們的運營支出項目、研發、銷售和營銷、一般及行政人員時，我們正在採取的行動以及我們採取的行動超過 40%-加上減少 - 人數和自由裁量權，你可以算一下。我要說的另一件事——我在準備好的評論中提到了這一點，這也影響了股票補償元素，儘管這不會影響調整後的 EBITDA，但我們也預計會下降。

So that is kind of the composition of what gives us confidence that we'll achieve adjusted EBITDA profitability in Q4.

因此，這就是讓我們有信心在第四季度實現調整後的 EBITDA 盈利能力的組成部分。

Got it. And then I guess on the cumulative [restaurant] bookings backlog, I guess they grew 10% sequentially. Can you talk about what kind of specific relationships or customers really drove the kind of quarter-over-quarter growth there?

知道了。然後我猜在累積的 [餐廳] 預訂積壓中，我猜他們連續增長了 10%。你能談談什麼樣的具體關係或客戶真正推動了那裡的季度環比增長嗎？

Yes. So, (inaudible) announced they have over [30,000] restaurants. And one thing that's very unique about our SoundHound for Restaurants is that we could work with single-location restaurants. There are a few players that are in the space because of the huge amount of onboarding requirements, it could take kind of a month to [pay in] a model for a new menu. They can only go after brands that have like hundreds of locations, for example, but we could actually work with a restaurant that has a single location, because our onboarding is automated. You can ingest the menu and even if they have custom vocabulary under menu or feature cognition, it automatically gets trained and enhance to recognize those and they have complex menu structure, all of that can be automated. So that's a very green area for us to go after the longtail. And the total partnership is very key because they have -- they opened the door to tens of thousands of restaurants for us, and we partnered with their sales team as well. So we have a large number of salespeople that can sell our solution on our behalf.

是的。因此，（聽不清）宣布他們擁有超過 [30,000] 家餐廳。我們的 SoundHound for Restaurants 非常獨特的一件事是我們可以與單一地點的餐廳合作。由於大量的入職要求，有一些玩家進入了這個空間，可能需要一個月的時間來 [支付] 新菜單的模型。例如，他們只能追求擁有數百個地點的品牌，但我們實際上可以與只有一個地點的餐廳合作，因為我們的入職流程是自動化的。你可以攝取菜單，即使他們在菜單或特徵認知下有自定義詞彙，它也會自動接受訓練和增強以識別那些，並且他們有復雜的菜單結構，所有這些都可以自動化。所以這是我們追逐長尾的一個非常綠色的區域。全面的合作夥伴關係非常關鍵，因為他們已經——他們為我們打開了通往數万家餐廳的大門，我們也與他們的銷售團隊合作。所以我們有大量的銷售人員可以代表我們銷售我們的解決方案。

Perfect. That makes sense. And I guess I could have one follow-up, the new kind of Generative AI chat product. Is this going to be more like a chat bot? Is it going to be more voice driven? And I guess how do you plan on, I guess, monetizing that solution, and maybe the timeline for when you expect monetization to occur? Is this something that we should expect to contribute to revenues this year?

完美的。這就說得通了。而且我想我可以有一個後續產品，一種新型的生成人工智能聊天產品。這會更像一個聊天機器人嗎？它會更受語音驅動嗎？而且我猜你打算如何將這個解決方案貨幣化，也許還有你期望貨幣化發生的時間表？這是我們今年應該為收入做出的貢獻嗎？

Yes. So the demo you heard was voice based, so the user will hop into a mobile app and hear the response. And we believe voice really unlocks the power of Generative AI. So tightening is not as accompanying that voice, because you can be faster and interact more freely. So it will be voice based. There will be a tech input as well. But of course, we think it will be the prominent way, and that's one of our differentiators, we bring our own ASR, and I really think this is the first time we can deliver a voice assistance to the end users that they have been wanting for more than a decade.

是的。所以你聽到的演示是基於語音的，因此用戶將跳入移動應用程序並聽到響應。我們相信語音確實可以釋放生成式 AI 的力量。所以收緊不是伴隨那個聲音，因為你可以更快，更自由地互動。所以它將是基於語音的。也會有技術投入。但是，當然，我們認為這將是突出的方式，這是我們的差異化因素之一，我們帶來了自己的 ASR，我真的認為這是我們第一次可以向最終用戶提供他們一直想要的語音幫助十多年來。

So if you remember when Apple launched Siri, more than 10 years ago, there were ads on TV that made a big promise that you can go and have an open ended conversation with an assistant. Users loved the idea, they rushed to it, and they quickly can disappointed, because it was able to only handle a few things, like ask a question about weather and maybe stock prices, and play music, and set timers.

所以如果你還記得 Apple 在 10 多年前推出 Siri 時，電視上的廣告承諾你可以去和助手進行開放式對話。用戶喜歡這個想法，他們衝上去，但很快就會失望，因為它只能處理一些事情，比如詢問有關天氣和股票價格的問題、播放音樂和設置計時器。

So it was still better than not having it. So people used it and adopted it, but it wasn't what they had imagined. And now more than a decade later, you are finally able to deliver that experience that users have wanted. We know they want it, because we've been in this business for a long time, for many years, and we know that's what the users want. And very few companies can take advantage of this moment, because just having a voice interface with a large language model is not enough, because maybe you can ask them some questions to write you a poem or some long tailed questions, but if you ask them questions about real-time search queries like, how is the weather or sports scores or stock prices or setting a timer or calling contacts or sending messages or setting calendar, navigating to restaurants or addresses, they (inaudible).

所以還是比沒有好。所以人們使用它並採用它，但它不是他們想像的那樣。十多年後的現在，您終於能夠提供用戶想要的體驗。我們知道他們想要它，因為我們從事這項業務已經很長時間，很多年了，我們知道這就是用戶想要的。很少有公司能利用這一時刻，因為僅僅擁有一個帶有大型語言模型的語音界面是不夠的，因為也許你可以問他們一些問題來給你寫一首詩或一些長尾問題，但如果你問他們有關實時搜索查詢的問題，例如天氣或體育比分或股票價格如何或設置計時器或呼叫聯繫人或發送消息或設置日曆，導航到餐館或地址，他們（聽不清）。

So to really deliver this ultimate assistant experience, you need to have both, you need to have the Generative AI large language models, and you need to have the existing assistant domains that people demand, with a voice interface on top of it, and there's only a handful of companies that can take advantage of the situation and SoundHound is the most agile, because we are able to move pretty fast. So stay tuned for our announcement later this month. The question -- the rest of the questions you asked, we will answer those in our announcements.

所以要真正提供這種終極助手體驗，你需要兩者兼備，你需要擁有生成式 AI 大型語言模型，你需要擁有人們需要的現有助手領域，在其之上有一個語音界面，還有只有少數公司可以利用這種情況，而 SoundHound 是最敏捷的，因為我們能夠快速行動。請繼續關注我們本月晚些時候的公告。這個問題——你問的其他問題，我們將在公告中回答。

And Brett, the other thing just for us, the Generative AI is certainly like the demo we shared, but we also talked about dynamic interaction and we previously announced the dynamic interaction sort of next generation for auto. And there's a kind of interest there in the space, we think certainly voice is an important modality, but sort of the multimodal is just as exciting. Sometimes voice input, voice output makes a lot of sense. Sometimes it's voice and touch input and audio and visual output, and that's sort of what our dynamic interaction capability allows us to do real time. And then that's really where the future is going, it's just going very rapidly and so there's a ton of application use cases that we're very excited about. And the monetization will come with those near term.

布雷特，另一件事只對我們來說，生成人工智能當然就像我們分享的演示一樣，但我們也談到了動態交互，我們之前宣布了下一代汽車的動態交互類型。在這個領域有一種興趣，我們認為語音當然是一種重要的方式，但多模態也同樣令人興奮。有時語音輸入、語音輸出很有意義。有時它是語音和触摸輸入以及音頻和視頻輸出，這就是我們的動態交互功能允許我們實時進行的事情。然後這就是未來的發展方向，它發展得非常快，因此我們對大量的應用程序用例感到非常興奮。貨幣化將伴隨那些短期的。

Thank you all for participating in today's question-and-answer session. I would now like to turn the call back over to Mr. Keyvan Mohajer for any closing remarks.

感謝大家參加今天的問答環節。我現在想將電話轉回給 Keyvan Mohajer 先生，聽取任何結束語。

Thank you. When we founded SoundHound, we had a vision to see voice-enabled, voice AI in our lifetime, but we couldn't have fully imagine the vibrant environment of AI innovation we find ourselves in today. Our conversational AI comes to its prominence across industries. We are confident that as the only independent cross-industry voice AI platform, we are in a strong position to deliver SoundHound's best-in-class technology to even more global brands.

謝謝。當我們創立 SoundHound 時，我們的願景是在有生之年看到支持語音的語音 AI，但我們無法完全想像我們今天所處的充滿活力的 AI 創新環境。我們的對話式 AI 在各行各業都很突出。我們有信心，作為唯一獨立的跨行業語音人工智能平台，我們有能力將 SoundHound 一流的技術提供給更多全球品牌。

We continue to execute and 2022 was a pivotal year for introducing new products and building new partnerships and we are excited about where we are and where we are looking forward to a strong 2023.

我們將繼續執行，2022 年是推出新產品和建立新合作夥伴關係的關鍵一年，我們對我們所處的位置感到興奮，並期待著強勁的 2023 年。

With that, I thank you for joining us on the call today and look forward to speaking with you when we report our first quarter 2023 results. Thank you.

因此，我感謝您今天加入我們的電話會議，並期待在我們報告 2023 年第一季度業績時與您交談。謝謝。

Thank you all for participating. This concludes today's program. You may now disconnect.

謝謝大家的參與。今天的節目到此結束。您現在可以斷開連接。