Reference
The meeting AI words,
in plain language.
An A-Z of the terms you meet when you evaluate an AI notetaker, defined honestly and briefly. No jargon defended with more jargon.
Meeting AI has its own vocabulary, and most of it is simpler than it sounds. These are the terms that turn up when you evaluate a notetaker, defined in a sentence or two, in plain language. Where a term maps to something MeetriX does, we have linked it. Roadmap items are labeled as roadmap items.
A
Action item
A follow-up recorded from a meeting, ideally with an owner and date. MeetriX can include action-oriented notes in the generated record; review them before creating or updating tasks.
Adaptive card
A rich, structured card format that renders natively inside Microsoft Teams channels. When a meeting ends, MeetriX can post one containing the summary, action items, and key quotes. See integrations.
Agenda
The ordered list of topics a meeting is supposed to cover, ideally with owners and time boxes. A good agenda is the cheapest meeting improvement there is; try the free agenda builder.
ASR (automatic speech recognition)
The technology that converts spoken audio into written text, also called speech-to-text. It is the first stage of every MeetriX pipeline, and the one where Arabic dialect coverage matters most. See Arabic transcription.
Attendee list
The record of who was actually in a meeting, as opposed to who was invited. MeetriX reports include attendance automatically, which matters for formal minutes and quorum evidence.
C
Calendar auto-join
Connecting a calendar (Outlook or Google Calendar) so the notetaker bot joins scheduled meetings on its own, governed by join rules you set. See calendar auto-join.
Code-switching
Switching between languages mid-conversation, or mid-sentence: an Arabic discussion with English acronyms, numbers, and asides. It is how real GCC meetings sound, and it breaks engines that transcribe one language at a time. MeetriX handles both languages in a single meeting.
Custom vocabulary
A user-defined list of canonical terms, such as product names, people, and company names, that the AI must keep exact in transcripts and summaries, including their Arabic and English transliterations. It stops "Vquity" from becoming "equity".
D
Data residency
A policy or legal requirement that certain data must stay inside a specific country or network. It is a common government constraint, and the reason MeetriX offers private-cloud and on-premise deployment. See security & on-prem.
Decision log
The recorded list of what a meeting actually decided, separate from what it discussed. MeetriX reports carry a dedicated decisions section so agreements do not dissolve into the transcript.
Dialect
A regional variety of a language with its own vocabulary, pronunciation, and grammar. Spoken Arabic is a family of dialects, Gulf, Levantine, Egyptian, and more, and MeetriX understands 32 of them plus English.
Diarization (speaker diarization)
Identifying who spoke when. MeetriX uses available participant channels or speaker-activity information to help attribute contributions. Room recordings and shared devices may require speaker-name review.
E
Email recap
A summary email sent automatically to attendees after a meeting ends. MeetriX recaps go out in Arabic or English, with or without speaker attribution, depending on your settings.
Executive summary
A short, high-level account of a meeting written for people who were not in the room: what happened, what was decided, what happens next. It is the first section of most MeetriX reports.
H
Hybrid meeting
A meeting where some people share a room and others join remotely. The remote side can be captured by the notetaker bot; the room side can be recorded in the browser or uploaded as audio. See in-person meetings.
I
In-person meeting
A meeting held physically, with no video call to join. MeetriX captures these too: record in the browser or upload an audio file, and the same transcription and report pipeline runs on it.
Integration
A connection between applications. MeetriX offers 840 integrations, with actions, permissions and deployment requirements configured for the chosen workflow. See integrations.
J
Join rules
Settings that select manually chosen, hosted or all calendar meetings; internal or internal-plus-external scope; and title-keyword inclusion and blocking. See calendar auto-join.
K
Keyword blocklist
A list of words that keeps the bot out of sensitive meetings: if a meeting title contains a blocked term, such as "1:1" or "legal", the bot never joins. Part of MeetriX join rules.
L
Language model
The AI that turns a raw transcript into structured output: summaries, minutes, decisions, and action items. In MeetriX it works in both Arabic and English and follows the report template you choose.
Late join
Arriving after the defined start time. Agree the cutoff, data source and reporting availability for your deployment before comparing punctuality. See meeting reporting.
Live captions
Real-time subtitles displayed while a call is happening. MeetriX does not do this today; it is on the roadmap. Today the transcript is available after the meeting.
M
Meeting intelligence
The product category MeetriX belongs to: capturing meetings, transcribing them, and turning them into searchable records, reports, and analytics, rather than just recording audio.
Meeting minutes
The formal, structured written record of a meeting: attendees, agenda, discussion, decisions, and actions. MeetriX generates minutes from the transcript, editable before they are shared, in Arabic or English.
MSA (Modern Standard Arabic)
The formal written register of Arabic used in news, documents, and official reports. Real meetings are mostly held in dialect, not MSA, which is why an engine trained mainly on MSA struggles in a real room.
N
Notetaker bot
A software participant that joins a video meeting to record and transcribe it. The MeetriX bot joins Zoom, Google Meet, Microsoft Teams, and Webex, appears in the participant list, and can be renamed and rebranded. See AI notetaker bot.
O
On-premise (on-prem) deployment
Running the agreed processing and storage components on the customer’s infrastructure. Online meeting capture and enabled external integrations have separate connectivity requirements. See deployment.
P
Per-speaker report
An individual recap generated for each attendee: what they said, what they committed to, and what concerns them. MeetriX can produce one for every person in the meeting.
Private cloud
A dedicated, single-tenant cloud environment under the customer's control, a middle option between shared cloud and fully on-premise. MeetriX supports all three.
Q
Quorum
The minimum number of members who must be present for a body to take valid decisions. Attendance records in formal minutes are how quorum is evidenced later, which is why MeetriX reports include who was actually there.
Quota
The number of meeting minutes a plan includes. MeetriX offers 600 free minutes to start, with upgrades on request. See pricing.
R
Report template
A layout for presenting a meeting record. Meeting Types define what to extract; Report Templates define how to arrange it. See meeting types and report formats.
RTL (right-to-left)
The writing direction of Arabic script. Proper RTL layout in transcripts, reports, and exports is table stakes for Arabic documents, and an afterthought in most notetakers.
S
Speaker attribution
Labeling every transcript line with the actual named person who said it, rather than an anonymous slot. It is what makes "who committed to what" answerable weeks later.
Speech-to-text
Another name for ASR: converting spoken audio into written text. See the ASR entry above.
Standup
A short, recurring team sync, often daily, covering progress, plans, and blockers. One of the most used MeetriX report templates.
Summarization level
How much detail a summary keeps. MeetriX offers two: key points only, for a fast scan, and detailed, for a fuller record. You can regenerate a report at the other level.
T
Talk time
The duration or share of a meeting in which someone speaks. Availability of automatic metrics depends on the deployment. The free talk-time analyzer lets you explore your own figures.
Text-to-speech
Generating spoken audio from written text, the reverse of ASR. On the MeetriX roadmap, not shipped today.
Transcript
The complete written record of who said what, line by line with timestamps. MeetriX transcripts are searchable, translatable between Arabic and English, and the source every report is built from.
Translation
Converting text between languages while keeping its meaning. MeetriX translates transcripts and summaries between Arabic and English, so one meeting can produce records in both. See translation.
Transliteration
Writing a word from one language in the script of another, such as an English product name spelled in Arabic letters. MeetriX custom vocabulary keeps both forms of your terms consistent.
V
VAD (voice activity detection)
Detecting which parts of an audio stream contain speech and which are silence or noise. It runs before ASR so the engine transcribes talk, not air conditioning.
W
WER (word error rate)
A speech-recognition metric that counts word substitutions, deletions and insertions relative to a reference. The published M3 evaluation reports character error rate (CER), which is a different metric. See M3 research.
White-labeling
Presenting the notetaker under your own brand: custom bot name, avatar, logo, and colors, so participants see your organization's assistant, not ours. An enterprise option. See teams & admin.
Workspace
A shared organisational context with members and managers. Configure record access and delivery to suit the team. See teams and administration.
Definitions are general explanations of how these concepts are commonly used. For how they work inside MeetriX, start with the transcription engine, walk the product tour, or learn the craft side in the Meeting Academy.
Frequently asked questions
What is the difference between ASR and speech-to-text?
Nothing: they are two names for the same thing, converting spoken audio into written text. ASR (automatic speech recognition) is the research term; speech-to-text is the product term.
What is speaker diarization, and why do names matter?
Diarization splits audio into "who spoke when". MeetriX uses available participant channels or speaker-activity information to help attribute contributions. Identity information depends on capture conditions and may need review.
What is a good word error rate (WER)?
Lower is better: 5% WER describes five word-level edits per hundred reference words. M3 reports 1.62% Arabic CER and 1.83% English CER in Lisan’s separate September 2026 evaluations. CER is not WER; read the scoring method at /research/m3/.
What is code-switching?
Switching languages mid-conversation or mid-sentence, such as an Arabic discussion with English terms, figures, and asides. It is how real GCC business meetings sound, and MeetriX transcribes both languages inside one meeting.
Does MeetriX show live captions during the call?
No, not today. Live in-call captions are on the roadmap. Today the bot records during the meeting and the transcript and reports are prepared after processing. See the notetaker bot.
Make your next meeting count.
Start with your next conversation. Or bring a representative meeting and let us help you plan an enterprise pilot.
Arabic and English · Cloud and private deployment · By Lisan