Dessa AI 推特 · X (Twitter) 熱門推文與學界評測精選
追蹤全網對 Block Dessa AI 的即時反響——從 Jack Dorsey 極簡的三詞力挺、專案負責人 Ragavan Thurairatnam 關於「音訊領域 GPT-3 時刻」的長文，到 GB200 單機櫃訓練與 M4 Max 本地推理的硬核解析。
@jack · Block Head & Cofounder, Block, Inc.
“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. block.xyz/dessa”
Block 共同創辦人 Jack Dorsey 官宣貼文：「我們相信聲音是與智慧互動最自然的方式，因此打造了迄今最真實的全雙工語音模型 Dessa。與它交談就像與真人對話……充滿情緒、細膩語氣、停頓與自然韻律。」
@RagavanThurair · Dessa AI Lead at Block (Former Cofounder, Dessa Toronto)
“We made the first AI model you can have a realistic conversation with that was trained entirely on speech! Dessa is a full-duplex speech-to-speech model that you can run on consumer GPUs. This is a GPT-3 moment for speech.”
Dessa 專案負責人 Ragavan Thurairatnam 推文：「我們打造了全球首個完全僅靠語音訓練、能進行真實即時對話的 AI 模型！Dessa 是可在消費級硬體執行的全雙工語音模型——這是屬於語音的 GPT-3 時刻。」
@RagavanThurair · Dessa AI Lead at Block
“What happens when you put Dessa on the phone with GPT-Live-1? Nobody steers it. Listen to how Dessa drops natural backchannels ('okay', 'right', 'that's relieving') while GPT-Live delivers textbook paragraphs.”
當讓 Dessa 直接與 GPT-Live-1 通話會發生什麼？全程無人工干預：聽聽 GPT-Live 在背誦教科書式段落時，Dessa 如何像真人一樣在換氣間隙自然附和（「okay」、「right」、「那我就放心了」）。
@RagavanThurair · Block Deep Learning Lab Engineering Note
“Trained from scratch on a single NVIDIA GB200 NVL72 rack. Runs locally faster than real-time on a MacBook Pro M4 Max or better. Zero text tokens. Hundreds of times fewer words than text LLMs.”
X 與 Hacker News 工程師社群最震撼的硬體指標：僅用單台 NVIDIA GB200 NVL72 機櫃從零完成訓練，詞彙量比文字大模型少數百倍，且能在 MacBook Pro M4 Max 上超即時本機推論。
@RagavanThurair · Acoustic Representation & Prosody Breakdown
“Say 'I didn't say she stole my CUDA code' out loud, stressing a different word each time. Same 7 words, 7 completely different meanings. Run speech through STT -> Text LLM -> TTS and all of that acoustic truth is destroyed.”
X 上瘋傳的「CUDA 程式碼重音測試」：大聲讀出「I didn't say she stole my CUDA code」，每次重讀不同單字，同樣一句話會產生 7 種截然不同的含義——而傳統語音轉文字（STT）管線會把這些聲學弦外之音徹底抹除。
@RagavanThurair · Early Access Call for AI Safety Researchers
“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. Calling all safety researchers!”
為什麼 Block 優先向 AI 安全研究員開放早期存取？「試想當你處於劇烈痛苦中，語音 AI 卻因為聽不出聲音裡的顫抖而自顧自說下去——這就是安全隱憂。」Dessa 團隊正公開招募安全學者進行測試。
@RagavanThurair · Unsteered Audio Sample Reaction
“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.”
X 上最受討論的實測細節：傳統語音模型會等你把話說完再按指令乾笑，而 Dessa 會在好笑的瞬間直接笑場；你讓它「大聲一點」，它就會自然提高音量——沒有任何人人工編寫過這條規則。
@RagavanThurair · AI-to-AI Collision & Turn Recovery Analysis
“In the Noisy GPT Live sample at 4:07, both AIs speak at the same time. GPT-Live-1 triggers a hard stop ('Sorry, go on.'), while Dessa smoothly replies: 'Oh, no, that's perfectly fine. I like what you're saying...'”
X 技術社群對《Noisy GPT Live》第 4 分 07 秒「搶話碰撞」的拆解：當兩個 AI 同時開口時，GPT-Live-1 觸發硬中斷生硬喊停（「Sorry, go on.」），而 Dessa 卻像真人一樣自然接話：「喔沒事沒事，我很喜歡你剛才說的……」
專案負責人 Ragavan Thurairatnam 證明：僅靠在原始人類對話音訊上做自迴歸預測，無需一行文字就能湧現出流利對話與推理能力。
作為直接向 @jack 匯報的神秘深度學習實驗室，Team Dessa 的首次公開亮相迅速引爆開源與端側 AI 社群。
開發者紛紛驚嘆：能在 MacBook Pro M4 Max 上脫離雲端、快於即時執行的全雙工語音模型，將徹底改寫語音助理的隱私與成本結構。
無論是吸氣聲、輕笑還是半句改口都逼真得令人起雞皮疙瘩，這也解釋了為何 Block 選擇先聯合安全學者評估風險再決定權重開源。