Dessa: Conversation Model trained only on speech
本文为原文前 6,000 字符的节选翻译，完整内容请查看原文。
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.
我们认为语音是与智能交互最自然的方式，因此我们打造了 Dessa，这是迄今为止最真实的全双工语音模型。与它交谈感觉就像在与真人对话……充满了情感、细微差别、停顿和自然的语调。
Interacting with current voice AI can just plain suck. It turns your voice into text and throws away an incredible amount of information text alone can’t carry. The most powerful AIs in the world can’t hear how you feel. We think that’s ridiculous, leads to a worse experience, and is potentially dangerous.
与当前的语音人工智能交互简直糟糕透顶。它将你的声音转化为文本，却丢弃了大量仅靠文本无法承载的信息。世界上最强大的人工智能也无法听出你的感受。我们认为这很荒谬，不仅导致体验变差，而且具有潜在的危险性。
We trained Dessa from scratch on nothing but human conversation. Sounds! No text, no pretrained models. It learned to hold a real-time conversation simply by listening. As far as we know, Dessa is the world’s first AI you can have a realistic live conversation with that was trained only from human speech.
我们完全从零开始，仅利用人类对话来训练 Dessa。全是声音！没有文本，没有预训练模型。它仅仅通过倾听就学会了进行实时对话。据我们所知，Dessa 是世界上第一个仅通过人类语音训练、且能与你进行真实实时对话的人工智能。
Dessa advances the frontier of deep learning models and creates a different category with entirely new and interesting behaviors for researchers to discover. How efficiently do they learn language and facts compared to text? How efficiently can they learn new languages? What sorts of In-context learning can they perform?
Dessa 推动了深度学习模型的前沿，并开创了一个全新的类别，为研究人员提供了全新的、有趣的探索行为。与文本相比，它们学习语言和事实的效率如何？它们学习新语言的效率如何？它们能进行什么样的上下文学习？
Every clip below is a live recording of a person talking to Dessa. Nobody steers it. Dessa picks what to say, when to say it and how, down to its own voice and personality. Subtitles were added afterward.
以下每一段剪辑都是一个人与 Dessa 交谈的现场录音。没有人引导它。Dessa 自己决定说什么、何时说以及如何说，甚至包括它自己的声音和个性。字幕是事后添加的。
It laughs the moment something’s funny, mid-sentence, the way people do. Other models wait for you to finish, then laugh on cue. Ask it to speak up and it gets louder. Nobody taught it that.
当事情变得有趣时，它会在句子中间立刻笑出来，就像人类一样。其他模型会等你讲完后再按指令发笑。要求它大声点，它的声音就会变大。没有人教过它这些。
The person blanks on a movie title. Dessa finishes the sentence. It also knows a surprising amount, considering every fact it has come from raw sound.
当对方忘记电影名称时，Dessa 会帮其补全句子。考虑到它掌握的每一个事实都来自原始声音，它所知道的信息量令人惊讶。
A surprise. Dessa gasps “oh my god” under its breath, then answers with what sounds like teary gratitude.
一个惊喜。Dessa 低声惊叹“天哪”，然后用听起来带着泪水的感激之情回答。
Huge range in intonation and speed.
语调和语速范围极广。
Here it sounds genuinely empathetic.
在这里，它听起来非常善解人意。
And here it talks to ChatGPT Live.
在这里，它正在与 ChatGPT Live 对话。
In this demo, Dessa is talking with GPT-Live-1, a third-party AI model. Both voices are AI-generated. The developer of GPT-Live-1 is not affiliated with Block and has not reviewed or endorsed Dessa.
在此演示中，Dessa 正在与第三方人工智能模型 GPT-Live-1 对话。两种声音均由人工智能生成。GPT-Live-1 的开发者与 Block 无关联，且未审查或认可 Dessa。
More recordings will be shared on X soon.
更多录音很快将在 X 上分享。
Audio in, audio out. Dessa is a true speech-to-speech model, and everything it knows came from human speech. It’s full-duplex, so it listens and talks at the same time, like you do. It learned from real conversation, so it knows when to cut in and when to backchannel (“yeah!”, “got it!”, “ooh”) to show it’s with you. Inference runs faster than real time, so there’s no awkward wait. And it runs locally on a MacBook Pro M4 Max or better.
音频输入，音频输出。Dessa 是一个真正的语音对语音模型，它所知道的一切都来自人类语音。它是全双工的，所以它能像你一样同时倾听和说话。它从真实的对话中学习，因此知道何时插话以及何时进行反馈（“是的！”、“明白了！”、“噢！”）以表明它在听。推理速度快于实时，因此没有尴尬的等待。它可以在 MacBook Pro M4 Max 或更高配置的设备上本地运行。
It natively hears and speaks emotion, sarcasm, laughter, a frustrated sigh, a voice cracking with sadness. Run speech through text and all of that is gone, which is why most voice AI feels stilted, slow and, worst of all, annoying.
它能原生听懂并表达情感、讽刺、笑声、沮丧的叹息以及因悲伤而颤抖的声音。如果通过文本处理语音，所有这些都会消失，这就是为什么大多数语音人工智能感觉生硬、缓慢，最糟糕的是，令人恼火。
LLMs need an insane amount of text. Dessa learned language from hundreds of times fewer words (in fairness, LLMs also have to learn C++ and whatever the hell Angular is).
大语言模型（LLM）需要海量的文本。Dessa 学习语言所用的词汇量要少几百倍（公平地说，LLM 还必须学习 C++ 和 Angular 之类的东西）。
And it had to learn far more than language. Accents, pauses, breathing, sarcasm, humor, inflection, emotion, when to cut in: every one changes what someone means. Say “I didn’t say she stole my CUDA code” out loud, stressing a different word each time. Same words, different meaning every time.
它必须学习的远不止语言。口音、停顿、呼吸、讽刺、幽默、语调、情感、何时插话：每一个因素都会改变说话者的含义。大声说出“我没说她偷了我的 CUDA 代码”，每次强调不同的词。同样的词，每次含义都不同。
Speech never repeats. Say a sentence twice, exactly the same way, and the data still comes out different. Text is identical no matter who types it. And a single spoken word takes thousands of times more bytes than the same word in text. People learn to talk so easily that we forget how hard it is. For a machine with no text to lean on, it’s brutal.
语音从不重复。即使以完全相同的方式说两次句子，数据结果依然不同。文本无论由谁输入都是一样的。而且，一个口语单词所占的字节数是文本中相同单词的数千倍。人类学习说话太容易了，以至于我们忘记了这有多难。对于一个没有文本可依赖的机器来说，这非常残酷。
But, we figured it out. And we figured out how to train Dessa on only one NVIDIA GB200 NVL72.
但是，我们做到了。而且我们找到了仅用一台 NVIDIA GB200 NVL72 训练 Dessa 的方法。
Three reasons: curiosity (aka fun), better understanding between humans and AI, and safety. Babies learn to talk without ever seeing a word written down. We wanted to see if a model could too. It can. It’s just very hard.
三个原因：好奇心（即乐趣）、人类与人工智能之间更好的理解，以及安全性。婴儿在从未见过书面文字的情况下学会了说话。我们想看看模型是否也能做到。它可以。只是非常困难。
Speech is the most natural interface to any intelligence, super or otherwise, and today’s voice AI wastes it. It flattens what we say into text and can’t keep up with human timing.
语音是与任何智能（无论是超级智能还是其他智能）交互最自然的界面，而当今的语音人工智能浪费了这一点。它将我们所说的话扁平化为文本，且无法跟上人类的节奏。
AI keeps gaining power and independence, and it’s still nearly deaf to the emotion in our voices. Imagine being in serious pain while a voice AI has no idea and just keeps going. That’s the safety problem. Dessa is a step toward fixing it.
人工智能不断获得力量和独立性，但对我们声音中的情感几乎充耳不闻。想象一下，当你处于剧烈疼痛中时，语音人工智能却毫无察觉并继续自顾自地说话。这就是安全问题。Dessa 是解决这一问题的一步。
We’re a (very) small AI research lab at Block, founded by former members of Dessa, the Toronto deep learning startup. We’ve built deep learning models since 2012 and generative speech models for over eight years. We chase frontier problems others ignore, and we report directly to Jack Dorsey, Block’s cofounder, Chairman and Block Head.
我们是 Block 内部一个（非常）小的人工智能研究实验室，由多伦多深度学习初创公司 Dessa 的前成员创立。自 2012 年以来，我们一直在构建深度学习模型，并从事生成式语音模型研究超过八年。我们追求其他人忽视的前沿问题，并直接向 Block 的联合创始人、董事长兼 Block Head Jack Dorsey 汇报。
You can reach us here.
你可以通过这里联系我们。
A model that learned to talk from sound alone is new territory, and we want outside eyes on it. We’re giving a small group of AI safety researchers early access to find the risks we haven’t anticipated. If that’s you, get in touch.
一个仅通过声音学会说话的模型是一个新领域，我们希望有外部视角来审视它。我们正在为一小部分人工智能安全研究人员提供早期访问权限，以发现我们未曾预料到的风险。如果你是其中一员，请与我们联系。
— Ragavan Thurairatnam, Dessa lead
— Ragavan Thurairatnam，Dessa 负责人
*in alphabetical order
*按字母顺序排列
In this demo, Dessa is talking with GPT-Live-1, a third-party AI model. Both voices are AI-generated. The developer of GPT-Live-1 is not affiliated with Block and has not reviewed or endorsed Dessa.
在此演示中，Dessa 正在与第三方人工智能模型 GPT-Live-1 对话。两种声音均由人工智能生成。GPT-Live-1 的开发者与 Block 无关联，且未审查或认可 Dessa。
Use your university, company, or lab email.
请使用你的大学、公司或实验室邮箱。
Loading application form…
正在加载申请表……
Effective Date: October 21, 2026
生效日期：2026 年 10 月 21 日
This Privacy Notice describes how Block, Inc. and our affiliates (collectively, “Block,” “we,” and “us”) collect, use, disclose, transfer, store, retain or otherwise process your information when you interact with the dessa.com website (the “Services”). This Privacy Notice applies only to information collected in connection with your access to and use of our Services. Please read this Privacy Notice carefully. By continuing to interact with our Services, you are consenting to the practices described in this Privacy Notice.
本隐私声明描述了当你与 dessa.com 网站（“服务”）交互时，Block, Inc. 及其关联公司（统称为“Block”、“我们”）如何收集、使用、披露、传输、存储、保留或以其他方式处理你的信息。本隐私声明仅适用于与你访问和使用我们服务相关的信息。请仔细阅读本隐私声明。继续与我们的服务交互，即表示你同意本隐私声明中描述的做法。
This privacy notice does not cover the processing of your informatio
本隐私声明不涵盖对你信息的处理