Analysing recordings in Webex, MS Teams and Zoom

Companies that rely on UC (unified communications) software for video meetings, especially in these times, are faced with the challenge of enforcing compliance policy and maintaining a record of those for legal or internal purposes.
This issue has been relatively easy to address for messages (as we have currently been doing for some time), emails, sms and any kind of written information.
While written information can be easily organized, classified and analysed, not is the case with audio and video recordings.

As we’re talking about companies with several meetings taking place daily and video conferences being preferred over chat and email, the only way to address it is via software, and more specifically, about machine learning models that can “understand” audio and video.

The Challenge

This poses an enormous amount of challenges. The main ones are getting a reliable transcription of what was said and identifying text and objects that are shown on screen in the video . Other challenges are  understanding the language of the conversation, understanding if there is usage of proper or common nouns, getting an algorithm that can distinguish between speakers in a conversation… and the list goes on and on.

As a response to a vast number of requests on this topic, AGAT Software has launched its solution for Compliance AI Recording Analysis. AGAT is bringing audio and video (including screen sharing) analysis for DLP and eDiscovery needs.

How does it work

This is a simple diagram of how the solution works

master recording infographics1 04 pegyi7
Diagram of Sphereshield Recording AI Compliance Analysis

Any DLP policy set to detect sensitive data is enforced on all communication types- messages, files, audio and video. , Once SphereShield detects such an incident an alert is created and awareness and preventive action is taken .

Watch the demo

The video shows 3 different scenarios implementing the Compliance AI Recording Analysis and how the user interface allows for search.

  • Sharing on screen an Excel with patient data (Social Security Numbers)
  • Discussing on the audio patient data
  • Sharing on screen a document which as a confidential classification mark on it

SphereShield can help addressing regulations such as FINRA, GDPR, HIPAA, MiFID II, PCI-DSS, amongst others

Main features:

  • Transcribe all recorded audio to 70+ languages
  • Analyse on-screen video with OCR
  • Search all meeting content using eDiscovery
  • Inspect meeting audio and video content by DLP policies
  • Ethical Wall – Enforce company policy on meeting attendee
  • On premises Archiving
  • Advanced AI and NLP Analytics

Transcribe all recorded meetings audio

Support audio transcription for 70 languages
Automatic language detection and Speaker enumeration: which speaker spoke which words and when.

Search meeting content using eDiscovery

Dates, Attendees names and type (internal/external), Text, Audio, Video, Keywords, Labels, Names, Etc.


-Inspect meeting audio and video for DLP policy compliance.
-Use SphereShield built-in DLP engine or integrate with leading vendors.
-Works with Symantec, McAfee, Google, GTB and Forcepoint.
-Smart DLP incidents with link to video and transcript time of violation

Ethical Wall

Validate meeting attendee using configurable compliance policies


Option to store video archiving on prem or on VPS

Advanced AI and NLP Analytics

for detection of:
-Named entities (products, countries, etc.)
-Emotion detection: Extract from what’s being said and voice tonality emotions such as joy, sadness, anger, or fear.
-Sentiments detection – positive / negative
-Automatic video scenes generation
-Labels and Keywords

More information by vendor:

Microsoft Teams Webex Zoom