# Browse longitudinal speaker profiles

Open the **Speakers** directory to review and edit profiles linked across managed transcripts (when you create or attach people during identification).

## Outcome

You will have opened **Speakers**, found a longitudinal profile in the people gallery, opened the dossier, and edited identity or inspected appearances.

## Starting point

Prefer completing [Identify and name speakers](speaker-identification.md) and choosing **Create new profile** or **Attach to** an existing person at least once. For a quick look, any existing profile in the Speakers directory is enough. Local AI is not required.

## What you’ll do

1. Open **Speakers**.
2. Confirm profiles are listed (not the empty state).
3. Search or open a card, then use the Identity and Appearances tabs.
4. Know when to return to Speaker Identification vs Settings → Speakers for voice batch.

## Walkthrough

1. Open **Speakers** from the primary sidebar section.

2. If you see **No speaker profiles yet**, go back to **Speaker Identification**, name a speaker, and choose **Create new profile** (or attach to an existing person), then return here. Settings → Speakers controls inclusion / enrol caps and library-wide voice batch — see [Settings & knobs](../runtime/settings.md#speakers).

3. When profiles exist, search by name or alias if needed, then open a card (for example a facilitator named during walkthrough 2). The header shows the display name and headline totals. Use **Identity** to edit name, aliases, notes, colour, or photo. Use **Appearances** to see linked transcripts and unlink if needed.

4. Use Speakers for cross-transcript continuity; use **Speaker Identification** for naming on a single transcript before analysis. Library-wide **Enrol trusted voice for all profiles** and **Pre-load voice suggestions** live on Settings → Speakers, not only the dossier Voice tab — see [Assist naming with voice](speaker-voice-matching.md).

5. After linking, re-open Speakers to confirm the profile still resolves — this is the durable directory, not a one-shot wizard result.

## What to notice

- Speakers profiles are longitudinal; diarized `SPEAKER_00` labels on a transcript are local until named and optionally linked. Ingest auto-link creates `auto_identified` links (review on Speaker Identification; promote before those appearances enrol as trusted voice) — [auto-identify.md](../runtime/auto-identify.md).
- Empty Speakers is normal on a brand-new library — create links from Speaker Identification rather than inventing profiles here.
- Privacy-sensitive voice evidence (when enabled) follows Speakers / Settings policy — keep disposable data roots for experiments.

## You should now have…

Opened the Speakers directory, located at least one longitudinal profile, and edited or inspected its identity (or a clear path to create the first link).

## Next

- [Identify and name speakers](speaker-identification.md)
- [Assist naming with voice](speaker-voice-matching.md)
- [Bundle transcripts into a group](groups.md)
- [Correct wording while reading](corrections.md)