Speaker Identification

Overview

AskElephant separates every recording into individual speakers and tries to match each one to a person in your meeting. How well it names those speakers depends on how the recording was made. This article covers what to expect from each recording method and how to get the best results.

How Speaker Identification Works

Desktop, mobile, browser recordings, and manual uploads all share the same process:

1. The recording is split into separate speaker lanes, one per distinct voice (a step called diarization).

2. AskElephant matches those lanes to the meeting's participants automatically.

3. Voiceprint matching can identify known voices when it's enabled and there's enough audio.

4. Anyone who can't be matched stays labeled Unknown Speaker.

5. You can correct speaker assignments manually at any time.

These methods don't receive speaker names from the meeting platform automatically. That behavior is specific to the AskElephant notetaker, covered below.

Desktop App

• Records your microphone and system audio together.

• The uploaded recording is separated into speakers after recording ends.

• The desktop app may show a live transcript with placeholder speaker labels, but this preview isn't used for the saved transcript.

• The participant list usually contains only you unless the recording is linked to an existing calendar meeting.

• The upload format supports participant details, but the current recording flow doesn't usually add them.

• Mixing microphone and system audio can make speaker separation less reliable.

Expected result: Speakers are separated, but names may stay unknown unless the recording is linked to a calendar meeting, voiceprints match, or you assign them manually.

Mobile App

• Records nearby audio through your phone's microphone.

• You can select meeting participants in the mobile recording flow.

• The participants you select are uploaded with the recording and become candidates for automatic matching.

• A recording started from an existing meeting can also inherit that meeting's participants.

• Mobile recordings don't use speaker information from the meeting platform.

Expected result: Speaker naming is usually better when you select attendees before uploading. Without selected participants, names may stay unknown.

Web App Recording

This is the browser's Record feature, not the notetaker.

• Records the microphone on the device running the browser.

• The browser recording flow doesn't currently include a participant picker.

• The participant list normally contains only the person who made the recording.

• Multiple voices can still be separated, but automatic matching has few identities to choose from.

Expected result: Multiple speakers may be detected, but most non-owner speakers are likely to stay unknown unless voiceprints match or you correct them manually.

Manual Meeting Upload

• The upload dialog lets you select participants for a new meeting upload.

• The participants you select become candidates for automatic matching.

• If you add a file to an existing meeting, AskElephant uses that meeting's existing participant list.

• The uploaded audio is separated into speakers after upload. File uploads don't contain speaker names from the meeting platform.

Expected result: This can produce strong results when you select the correct participants. Uploading without participants behaves more like a browser recording.

How the Notetaker Is Different

A meeting recorded by the AskElephant notetaker works differently:

• It combines speaker separation with the meeting platform's active-speaker information.

• Speakers can start with names captured from Zoom, Google Meet, or Teams.

• AskElephant then matches those names against your calendar attendees and known contacts.

• Automatic matching and voiceprints stay as fallback and enrichment methods.

This is usually the strongest source for automatic speaker naming.

Which Methods Name Speakers Best

From most to least identity context:

1. Notetaker recording.

2. Mobile recording with attendees selected.

3. Manual upload with attendees selected.

4. Desktop recording linked to a calendar meeting.

5. Desktop or browser recording without participant information.

Voiceprint matches and manual corrections can improve any of these results.

Need Additional Help?

If you have questions or need further assistance, the AskElephant support team is here to help!

You can reach our support team in several ways:

• click the Support tab on the right side of your screen,

• email us at [email protected],

• or use @askelephant support in your dedicated Slack channel.

We're committed to getting you the answers you need as quickly as possible.