Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.wittify.ai/llms.txt

Use this file to discover all available pages before exploring further.

Voice Library is a dedicated page that lists every custom voice you have cloned or designed. The page subtitle reads Your cloned and designed voices live here. Use them from the Studio’s voice picker. Voices are private to your account, only you can see them.

Page anatomy

The page is one full-width column inside the dashboard.
RegionWhat you see
HeaderTitle Voice Library, the subtitle, and a small voice count line directly under it that reads voices (or 1 voice in the singular case).
Create A Voice buttonA brand-gradient pill on the end edge of the header with a + icon. Opens the Create a new voice picker.
Voice gridA responsive grid of voice cards (1, 2, or 3 columns depending on viewport). Empty when you have not made any voices yet.
Empty stateA dashed-border card centered on the page with a microphone icon, the heading No custom voices yet, the body Create your first custom voice by cloning from a sample or designing one from attributes., and a Create your first voice button.
Topbar (provided by the dashboard)A second + Create Voice button. Clicking it opens the same picker as the header button.

The voice card

Each card is split into a 16:9 cover area and a body.

Cover area

ElementNotes
Cover imageThe image you uploaded when you cloned or designed the voice. The image is cropped to fill the cover.
Fallback tileWhen you did not upload a cover, a 64 px round tile takes the cover space, showing the voice’s first letter. The tile color is generated deterministically from the voice id, so the same voice always shows the same tint.

Body

ElementWhat it says
NameThe name you gave the voice when you created it.
Status chipTop-end-edge corner of the body. Ready (green), Training (amber), or Failed (red).
Source pillA small rounded chip showing the source. Cloned with a microphone icon if you uploaded or recorded a sample. Designed with a magic-wand icon if you described one from attributes.
Language codeThe macro language code from creation, in mono digits, left to right.
Dialect codeWhen set, the dialect code appears next to the language, separated by a middle dot.
Transcription previewWhen the voice was cloned and the transcription was confirmed, the first two lines of that transcription appear below the chips.
Audio sampleWhen the voice has a saved sample, an inline browser audio player lets you scrub and play.
Delete actionThe end edge of the card footer carries a small icon button with a trash icon and the label Delete. Hovering it tints to the destructive (red) color.

Status pipeline

A voice you just created starts in Training. The card stays on the page while the engine processes the sample or attributes, and the chip flips to Ready when training succeeds, or Failed when it errored.
ChipMeaningWhat you can do
ReadyThe voice is searchable in the Voice Picker and usable in Studio.Pick it in any future generation.
TrainingThe engine is still processing.Wait. The card refreshes by itself when the status changes.
FailedThe engine reported an error.Delete and try again. The most common cause is a sample that was too quiet or too short.
You can keep using the rest of the app while a voice trains. The card stays on the Voice Library page, you do not need to keep this page open.

Creating a voice

There is no inline create form on this page. Both the header Create A Voice button and the topbar + Create Voice button open the same picker. The Create a new voice picker is a small dialog with two cards.
CardTitleDescription
CloneClone a voiceUpload or record a short sample. We will match tone, accent, and pacing. Opens the Clone Wizard.
DesignDesign a voiceDescribe how it should sound, and we will generate it from scratch. Opens the Design Wizard.
Both wizards land the new voice on this page when they finish.

Deleting a voice

Click the Delete action in a card’s footer. A confirmation dialog appears.
FieldCopy
TitleDelete this voice?
DescriptionAnyone using this voice will lose access to it. You cannot undo this.
Cancel buttonCancel. Closes the dialog with no change.
Confirm buttonDelete. Styled with the destructive (red) background and white text. Removes the voice immediately.
After confirming, a toast surfaces the message Removed from history. and the card disappears from the grid.
Deletion is permanent for that custom voice. Every History project that referenced this voice keeps its audio (already generated speech does not vanish), but you cannot generate new speech from it. There is no restore.

Tenant isolation

Every voice is tied to your account. Other users in the same workspace cannot see your custom voices, and you cannot see theirs. Sharing a voice across accounts is not supported.
If you ever see another user’s voice in your library, that is a bug — please report it.

Provider routing

Voices are Wittify voices. The underlying provider (the actual model that produces the audio) is internal routing detail and never appears on the cards. You will never see provider names in the Status, Source, or any other field.
This is a hard product rule: never surface provider names anywhere in the Voice Library. The card’s color tint is derived from the voice id, not the provider.

Common questions

Not from this page. The voice keeps the name you gave it in the wizard. Delete and recreate if the name is critical. The cloned audio sample is the slow part, the wizard remembers most of your choices.
Not yet. Cross-account voice sharing is not supported. Each user creates their own private library.
Anywhere there is a voice picker: the Studio’s voice card, every wizard step that lets you pick a voice, and the My Voices tab in the Voice Picker modal.
The fastest path is to delete the failed card and run the wizard again. Most failures come from samples that were too quiet, had heavy background noise, or were shorter than the minimum length (3 seconds for cloning).
Check the My Voices tab in the picker. The picker uses tabs to separate built-in voices from yours. Voices in Training still show in My Voices but cannot be selected for a generation until they reach Ready.
For cloned voices, the sample is the audio you uploaded or recorded. If it is rough, the cloned voice will be too. Re-record from a quieter room, hold the mic 10 to 15 cm away from your mouth, and aim for a 3 to 10 second clip without long silences.
Look for the dashed-border card in the middle of the page. The button inside reads Create your first voice. The header Create A Voice button on the end edge of the page does the same thing.

Where to go next

Voice Picker

Browse all voices including your custom ones.

Clone Wizard

Add a cloned voice.

Design Wizard

Add a designed voice.

Studio

Use a custom voice in a generation.