Introduction
You just walked out of a two-hour project review. Decisions were made, action items assigned, and three people said "let's circle back" — but nobody wrote any of it down. This is the problem an AI meeting assistant solves: capture every word, separate who said what, and hand you structured meeting notes without anyone breaking eye contact to type.
Two very different products fit this brief. MeetingMinutes is a standalone mobile-first AI meeting recorder built around on-site capture, dialect recognition, and one-click export. Feishu Miaoji (飞书妙记), known as Lark Minutes internationally, is the transcription layer embedded inside the Feishu/Lark collaboration suite. Both turn speech into searchable text and produce AI summaries — but they were built for different working styles.
What to Look for in an AI Meeting Assistant
Anchor on three criteria that matter in real use:
Capture flexibility — can it record in-person meetings offline, import existing audio files, and capture online calls?
Transcription quality — accuracy under real-world noise, speaker separation, and handling of dialects or second languages.
Post-meeting output — a usable transcript, a shareable summary, and export formats that fit your team's tools.
MeetingMinutes: Built for On-Site Capture and Multi-Source Import
MeetingMinutes is an AI meeting assistant available as a mobile app with a companion web client. Its design centers on recording real-world meetings where people are in the same room — or where the audio already exists elsewhere.
Three features drive its value:
High-precision real-time transcription — standard Mandarin reaches up to 98% accuracy, and the engine strips filler words and repeated pauses so the transcript is readable on first pass. Real-time transcription lets you follow along in medium conference rooms and offline lectures.
Dialect and multilingual coverage — it supports over 20 Chinese dialects (Cantonese, Sichuanese, Shanghainese, Hunanese, and others) alongside 52 languages with native-accent recognition. This suits regional field interviews, supplier visits, and mixed-language internal calls.
Offline recording and broad import — a local recording engine saves audio even with no network, critical for government on-site meetings or fieldwork. It also accepts 9 audio and 13 video formats, extracts audio from short-video links (Douyin, Bilibili, Xiaohongshu), and imports WeChat voice messages directly.
After recording, MeetingMinutes generates AI summaries from 50+ templates covering medicine, law, and interviews, plus mind maps, slide decks, and Excel exports. Files sync across mobile and web, and can be shared via QR code, WeChat, DingTalk, QQ, or email.
Feishu Miaoji (Lark Minutes): Transcription Inside the Collaboration Suite
Feishu Miaoji is an AI meeting assistant embedded in the Feishu/Lark workplace. It is not a standalone recorder; its power comes from being connected to Feishu video meetings, Docs, Tasks, and the company directory.
Key capabilities confirmed by Feishu's public documentation:
Meeting capture — during a Feishu video call, the host can enable AI summary or recording; minutes are generated automatically afterward. It also supports mobile audio recording and uploading local audio/video files. A single recording can run up to 24 hours.
Transcription and speaker ID — real-time transcription with speaker separation; if members enrolled their voiceprint, the transcript shows their actual names. Feishu reports ASR accuracy above 98% for standard Mandarin.
AI meeting notes — post-meeting minutes are structured into key decisions, pending topics, and information syncs, with auto-detected action items pushed into Feishu Tasks. In-meeting AI summary can run live during the call.
Multilingual and collaboration — meeting minutes translate between languages in one click, and real-time translation supports Chinese, English, and Japanese. Colleagues can comment, @mention, and edit minutes inside Feishu.
Pricing — the free Feishu Basic plan includes 300 transcription minutes per month. Business and Enterprise plans offer unlimited Minutes transcription (subject to organization storage).
Side-by-Side Comparison
| Feature | MeetingMinutes | Feishu Miaoji (Lark Minutes) |
|---|---|---|
| Real-time transcription | Yes, mobile-first | Yes, in Feishu calls and mobile recording |
| In-person / offline recording | Local offline engine, no network needed | Mobile recording requires network for transcription |
| Online meeting capture | Mobile screen-recording import or audio extraction | Native inside Feishu video meetings |
| Speaker identification | AI voiceprint, numbered tags | Voiceprint ID shows names if enrolled |
| Language support | 52 languages + 20+ Chinese dialects | CN/EN/JP real-time; minutes translation |
| Audio/video import | 9 audio + 13 video formats; short-video links; WeChat audio | Local file upload; Feishu Docs import |
| AI summaries | 50+ templates; mind map, PPT, Excel output | Structured chapters + action items to Feishu Tasks |
| Team collaboration | QR code, WeChat/DingTalk/QQ/email sharing | Native Feishu comments, @mentions, Docs co-edit |
| Export options | Multiple file formats | Feishu Docs; share within Feishu |
| Storage | Cloud + local dual backup | Unified Feishu cloud storage |
| Free tier | Freemium mobile app | 300 transcription minutes/month on Basic |
Differences in Real-World Use Cases
In-Person and Off-Site Meetings
Walking a factory floor, sitting in a government hearing, or interviewing in a branch with weak signal: MeetingMinutes' local engine saves audio first and transcribes later — more reliable. Feishu Miaoji assumes a connected session, since minutes generation depends on Feishu's cloud.
Online Meetings Inside Your Company
For teams already on Feishu, Miaoji is smoother: recording starts inside the call, action items land in Feishu Tasks, and the minutes doc appears in the right chat group. No separate app.
Regional and Dialect-Heavy Interviews
When suppliers speak Cantonese, Sichuanese, or Shanghainese, MeetingMinutes' dialect models are built for this. Feishu's multilingual strength leans toward global team translation (CN/EN/JP) rather than Chinese dialects.
Post-Meeting Asset Production
MeetingMinutes turns one recording into many artifacts — mind maps, PPT, Excel, and a graphic summary — suiting consultants and journalists who repurpose audio. Feishu Miaoji optimizes for one high-quality shared doc inside the suite.
Where MeetingMinutes Stands Out
MeetingMinutes is strongest when the meeting happens outside the Feishu walled garden: off-site visits, dialect-heavy fieldwork, offline government meetings, or a backlog of audio from other platforms. Its broad import channels and template-based output make it a flexible post-production tool for solo knowledge workers and small teams.
Where Feishu Miaoji May Be a Better Fit
If your company already uses Feishu for chat, docs, and video, Miaoji removes the friction of a second tool. Action items auto-assign, minutes land in the meeting group, and permissions inherit from your org chart. Teams that value governance, directory-level speaker naming, and Feishu-native workflow will prefer it.
Which One Should You Choose?
Choose MeetingMinutes if you need:
Reliable offline or off-site recording in weak-signal environments
Chinese dialect recognition or 52-language transcription
Import from WeChat, short-video platforms, or mixed media files
Export to mind maps, PPT, Excel, or share outside Feishu
Choose Feishu Miaoji if you need:
Native transcription inside Feishu video meetings with auto action items
Minutes that flow into Feishu Docs, Tasks, and group chat
Enterprise administration and directory-based speaker naming
Transcription bundled into an existing Feishu/Business subscription
Final Takeaway
MeetingMinutes and Feishu Miaoji both turn spoken meetings into searchable, shareable AI meeting notes — but they sit at different points on the product spectrum. MeetingMinutes is a portable, import-friendly AI meeting recorder built for where the conversation actually happens, including offline and dialect-heavy settings. Feishu Miaoji is a suite-embedded transcription layer that shines when your team already lives in Feishu. The right choice comes down to whether your workflow needs to capture the world outside your chat app, or deepen what already happens inside it.