Wispr Flow 更新：自研语音模型 Canto 与 Windows 版 Notetaker
听写延迟降 14%、内存降 27%，新增自动语言识别与 Hinglish 支持；会议记录器登陆 Windows，支持实时翻译、Outlook 日历和 Gemini/Cursor MCP。
Team has been shipping. Here’s a rundown of everything that’s new: Dictation: > Shipped Canto, our first in-house speech model > Shipped an update that makes dictation use 27% less memory on Mac and Windows > Cut median dictation latency by 14%, so your words appear faster > Shipped automatic language detection across your selected languages > Shipped Hinglish dictation in Latin script > Added a prompt when you speak a language you haven’t selected > Added instant warnings when your microphone isn’t picking up sound > Added shared dictionary entries for your organization or department Notetaker: > Shipped Notetaker on Windows > Shipped live translation so you can follow meetings in your preferred language > Shipped an update that makes Notetaker use 35% less memory on Mac and Windows > Shipped folders to organize meeting notes and share them with your team > Added Outlook calendar support > Added Otter imports, bringing your previous transcripts into Notetaker with speaker names preserved > Added one-click Gemini+ Cursor MCP, so your AI tools can work with your meeting context > Shipped more detailed summaries that adapt to the length of your meeting > Made “Ask Anything” faster > Improved speaker detection in Teams gallery view and while presenting in Google Meet > Added @ mentions and contact cards in meeting notes > Improved recording and upload recovery, including automatic recovery for stuck Windows uploads We’re shipping every day. What do you want to see next? 🗣️