Rime 发布语音模型 Prism，先写文本再发声
Rime has launched Prism, its new speech-to-speech model for AI voice agents that writes each reply as text before saying it out loud.
Rime 发布面向 AI 语音智能体的语音到语音模型 Prism，它在开口前先把每条回复写成文本。企业可以编辑转写内容并拦截智能体不该说的话，Prism 同时保持通话不中断，并能在对话中查询账户或冻结卡片。Rime 称没有其他语音到语音模型采用这种方式，因此 Prism 更适合银行等受监管企业。
查看原帖隆重介绍 prism，全球首个在对话中让人分辨不出是人类的语音到语音模型 prism 拥有： > 300 毫秒响应时间，比 OpenAI GPT-Live 快 4 倍 > 在智能表现上超越所有其他 s2s 模型（且无需推理） > 首个可通过强化学习定制的 s2s 模型 我们是一支 35 人的团队，已获得 3500 万美元融资，我们推动 s2s 模型普及的速度比 openai、google 和 x 更快。 前 100 位在此注册的用户可免费使用 1 个月：https://t.co/KC3KUq18bl
查看原文
introducing prism, the world’s first speech-to-speech model you can’t tell from a human in conversation prism has: > 300ms response time, 4x faster than OpenAI GPT-Live > outperforms every other s2s model on intelligence (with NO reasoning) > the first s2s model customizable through reinforcement learning we’re a team of 35 with $35M in funding, and we’re unlocking s2s model adoption faster than openai, google, and x. the first 100 people to sign up here get 1 month free: https://t.co/KC3KUq18bl
正文
Rime has launched Prism, its new speech-to-speech model for AI voice agents that writes each reply as text before saying it out loud.
Companies can edit the transcript and block anything the agent shouldn't say, while Prism keeps the call going and looks up an account or freezes a card mid-conversation.
Rime says no other speech-to-speech model works this way, which makes Prism a safer fit for banks and other regulated businesses.