Glossary of meetings, transcription and data

What the terms that come up when recording, transcribing and subtitling meetings, and protecting the data of the people in them, actually mean.

A

Action items: Specific commitments that come out of a meeting: what needs doing, who does it and by when.
AI hallucination: An AI answer that sounds convincing but isn’t true or isn’t supported by the information it was given.
AI meeting summary: A summary of a meeting or recording written by a language model from the transcript: key points, decisions and actions.
Audit log: A dated list of who did what in an application: who uploaded, downloaded, shared or deleted what.
Automatic transcription: Transcription done by speech recognition software, without anyone listening to the recording and typing it out.

Burned-in subtitles: Subtitles that are part of the video picture and always visible, in any player, with no way to turn them off.

Custom vocabulary (glossary): A list of names, brands, acronyms and technical terms given to the system so it spells them correctly when transcribing.

Data controller: The person or organisation that decides why and how personal data is processed, and is accountable for it under the law.
Data processing agreement (DPA): The contract required by Art. 28 GDPR between controller and processor: which data, for what and with which guarantees.
Data processor: A person or organisation that processes personal data on behalf of another (the controller) and following its instructions.
Data retention: How long personal data is kept before it’s deleted. The GDPR says it shouldn’t be kept longer than its purpose requires.
Dubbing (synthetic voice): Replacing a video’s original voice with one in another language, recorded by actors or generated with a synthetic voice.

GDPR: The General Data Protection Regulation: the EU law governing how personal data is collected, used and stored.

Large language model (LLM): A generative AI model trained on huge amounts of text that predicts words to summarise, answer questions, translate or draft.
Lawful basis: The legal ground that allows personal data to be processed under Art. 6 GDPR, such as consent or legitimate interests.
Live transcription: Text that appears while someone is speaking, a few seconds behind, instead of once the recording has finished.

Machine translation: Translating text from one language to another with software, without a human translator.
MCP (Model Context Protocol): An open standard for connecting AI assistants to external applications and data through tools the assistant can use.
Meeting agenda: The list of topics a meeting will cover, in the planned order and often with a time for each one.
Meeting bot (AI meeting assistant): An automated participant that joins a video call to record and transcribe it, then delivers the summary afterwards.
Meeting follow-up: What happens after a meeting to make it count: sending the summary, sharing out actions and checking they get done.
Meeting minutes: A document recording who attended a meeting, what was discussed, what was decided and which actions are outstanding.
Minuta (meeting minutes in Latin America): In much of Latin America, the document that summarises a meeting: what Spain usually calls an «acta» and English calls minutes.

Pseudonymisation and anonymisation: Techniques so data doesn’t identify anyone directly (pseudonymise) or by any reasonable means at all (anonymise).

Speaker diarisation: Splitting a recording by who is speaking at each moment: Speaker 1, Speaker 2… without yet knowing their names.
Speaker recognition: Identifying which specific person a voice belongs to by comparing it with known voices. It isn’t the same as separating speakers.
Special category data: Especially sensitive personal data (health, biometrics, beliefs…) whose processing Art. 9 GDPR prohibits except in specific cases.
Speech recognition (ASR): Technology that turns spoken words into text, also known as ASR (automatic speech recognition) or speech-to-text.
SRT (subtitle file): A plain-text subtitle format: each block has a number, an in and out time and the text.
SSO (single sign-on): A system that lets people sign in to several applications with their company account, without a separate password for each.
Sub-processor: A supplier a data processor uses to process its clients’ personal data, with the controller’s authorisation.
Subtitle reading speed (characters per second): How many characters per second a viewer must read to follow a subtitle: too high and there isn’t time to read it.
Subtitles: Text synced with a video showing what’s said, in the same language or translated, in short blocks that appear and disappear.
Subtitles for the deaf and hard of hearing (SDH): Subtitles for deaf and hard-of-hearing viewers: as well as dialogue, they show who is speaking and relevant sounds.

Timestamps: Markers showing the exact point in the recording at which each word, sentence or paragraph of a transcript was said.
Transcription: Turning what is said in an audio or video recording into written text, word for word or with light tidying for readability.

Voiceprint: A numerical representation of a voice’s characteristics used to recognise the person. It counts as biometric data.
VTT (WebVTT): A subtitle format built for the web: similar to SRT, with a “WEBVTT” header and options for position and styling.

Webhook: An automatic notification one application sends to another’s web address when something happens, such as “the transcript is ready”.
WER (word error rate): The percentage of words an automatic transcript changes, drops or adds compared with the correct text. The lower, the better.

Try it with your next meeting.

Create your account in a minute or ask us for a demo for your team.