Tips & Tricks¶
Secrets of the Craft¶
A collection of power-user tips, hidden features, and best practices to get the most out of TaleKeeper.
Keyboard Shortcuts¶
| Key | Action |
|---|---|
| 1 | Switch to Recording tab |
| 2 | Switch to Chronicle (Transcript) tab |
| 3 | Switch to Tales (Summaries) tab |
| 4 | Switch to Visions (Illustrations) tab |
| 5 | Switch to Export tab |
| Enter | Save session name (while editing) |
| Esc | Cancel session name edit |
Note
Keyboard shortcuts are disabled when you're typing in a text box or using a dropdown menu.
Session Status¶
Sessions move through these stages automatically:
| Status | Meaning |
|---|---|
| Draft | Session created, no audio yet |
| Recording | Microphone is live |
| Audio Ready | Audio captured, awaiting processing |
| Transcribing | Converting speech to text and identifying speakers |
| Completed | Ready for summaries, illustrations, and export |
Auto-Resume Processing¶
If you navigate away from a session while it's processing (transcribing, diarizing, etc.), TaleKeeper automatically resumes tracking progress when you return. No need to restart — just come back to the session and the progress bar picks up where it left off.
Language Search by ISO Code¶
When selecting a language in any language dropdown, you can type an ISO language code (e.g., zh for Chinese, he for Hebrew, ja for Japanese) to jump directly to it. The dropdown searches both language names and codes.
Regenerate Session Name¶
Completed sessions show a Regenerate Name button next to the title. Click it to have the AI craft a new narrative title based on the transcript content — useful if the auto-generated name doesn't capture the session's essence.
Audio Keeps Playing Across Tabs¶
You can freely switch between the Recording, Chronicle, Tales, Visions, and Export tabs without interrupting audio playback. Your listening position is preserved — jump to Tales to check a summary, then switch back to Chronicle and the audio is right where you left it.
Write Your Own Image Prompts¶
You don't have to use Generate Scene before creating an illustration. Type any prompt you like directly into the text area on the Visions tab and click Generate Image. Great when you have a specific moment or character pose in mind.
Edit All Speakers at Once¶
Instead of editing speakers one by one, click Edit All in the speaker panel to open every speaker for editing simultaneously. Your roster members appear as quick-fill buttons — click one to instantly assign both player and character name.
Availability Warnings¶
The Tales and Visions tabs automatically check whether your LLM and image providers are reachable when you open them. If something isn't connected, a warning banner appears before you try to generate — saving you from waiting on a request that would fail.
Optimized for Group Recording¶
TaleKeeper's recording is tuned for multi-person tabletop sessions. Some automatic audio filtering that normally helps with one-on-one calls is turned off because it can distort recordings when several people speak around the same microphone.
Stuck Recording Lock¶
Only one session can record at a time. If you see "Another session is recording" but nothing is actually recording, simply close any old TaleKeeper browser tabs and try again.
Best Practices¶
Recording Tips:
- Set the speaker count to match your actual party size (including the DM) — it significantly improves speaker detection
- Record in a quiet environment when possible — background noise hurts both transcription and diarization
- Use an external microphone if available — laptop mics pick up more noise
Model Selection:
- Start with distil-large-v3 (the default) — it offers excellent accuracy at fast speeds on Apple Silicon
- Try large-v3 for important sessions, heavily accented speech, or mixed-language games
- Use tiny or base only for quick tests or if your machine struggles with larger models
Speaker Management:
- Name speakers early — before generating summaries, so the AI knows character names
- Fill in your roster with descriptions — this enriches both summaries and illustrations
- If diarization split one person into two speakers, merge them before generating summaries
Summaries & Illustrations:
- Longer sessions produce better summaries — the AI has more material to work with
- Generate illustrations after summaries — the AI uses summary content to craft better scenes
- Character descriptions matter — the more detail in your roster, the better your illustrations
Multi-Language Support¶
TaleKeeper supports 98 languages across the entire pipeline:
- Transcription — Whisper recognizes speech in all supported languages
- Summaries — generated in the session's configured language
- Session names — auto-generated in the correct language
- Scene descriptions — image prompts respect the source language context
Set the language at campaign level (applies to new sessions) or override per-session.
All Hidden Features at a Glance¶
| Feature | Where to Find It |
|---|---|
| Process All pipeline (one-click transcribe → diarize → summarize → illustrate) | Recording |
| Multi-part audio upload with ordering and merge | Uploading |
| Voice signature confidence threshold | Voice Signatures |
| D&D Beyond character import | Roster |
| PDF character sheet extraction | Roster |
| Generic URL character import | Roster |
| Sheet refresh / re-extraction | Roster |
| Session auto-renumbering | Campaigns |
| Speaker count override at recording end | Recording |
| Retranscription with different settings | Retranscription |
| Speaker merging | Merging |
| Re-diarization without re-transcription | Speakers |
| Voice signatures | Voice Signatures |
| POV character journal entries | Journals |
| Auto session naming | Summaries |
| Scene description editing | Illustrations |
| Character appearance in illustrations | Illustrations |
| Printable PDF mode | PDF Export |
| Batch ZIP export | PDF Export |
| Hero image in PDFs | PDF Export |
| Email sharing | |
| Continue Last Session quick access | Campaigns |
| Session content badges (Audio, Transcript, Summary, Images) | Campaigns |
| Inline session name editing | Campaigns |
| Regenerate session name with AI | Tips & Tricks |
| Transcript search and filter (text + speaker names) | Transcription |
| Copy transcript line to clipboard | Transcription |
| Bidirectional audio ↔ transcript sync | Transcription |
| Crosstalk segment indicators | Transcription |
| Voice sample upload from roster | Roster |
| Dark / light theme toggle | Settings |
| Reset to Defaults (preserves API keys) | Settings |
| Auto-resume processing after navigation | Tips & Tricks |
| Language search by ISO code | Tips & Tricks |
| Audio keeps playing across tabs | Tips & Tricks |
| Free-form image prompts (skip Generate Scene) | Tips & Tricks |
| Edit All speakers with roster quick-fill | Tips & Tricks |
| LLM / image availability auto-check | Tips & Tricks |
| Clearing a stuck recording lock | Tips & Tricks |
| Automatic speaker recognition via voice signatures | Voice Signatures |
| Upload voice sample before first session | Roster |
| Generate voice signatures from labeled session | Speakers |
| Edit All speakers at once | Speakers |
| Automatic segment splitting at speaker changes | Transcription |
| Volume normalization for quiet speakers | Transcription |
| Color-coded speaker labels | Transcription |