Voice library overview
Navigate to Voices in the sidebar. The main library page presents:Trending voices
The top 10 voices sorted by usage count across the ElevenLabs shared library.
Curated collections
Handpicked groupings: Professional Narrators, Character Voices, and News & Information.
Best for English
Top-rated English-language voices, sorted by popularity.
Language collections
Voices grouped by language (English, Spanish, French, German, and more). Each group shows how many voices are available.
My Voices
My Voices (sidebar sub-item) shows only the voices associated with your ElevenLabs account — that is, voices in thecloned category or voices you have added to your account. Customer isolation is applied: you only see voices that were created under your account prefix.
From this page you can:
- Preview a voice (proxied through Sniko to avoid CORS issues).
- Delete a cloned voice. Default (built-in) voices cannot be deleted.
- Search by name or description.
Default Voices
Default Voices shows the curated set of built-in ElevenLabs voices available to all accounts. These include well-known voices such as Rachel, Adam, Bella, Josh, and others. Default voices can be filtered by category and language but cannot be deleted.Instant Voice Clone
The Instant Voice Clone feature lets you create a custom voice from one or more audio recordings. The cloned voice is added to your ElevenLabs account and immediately appears under My Voices.Requirements
Before uploading, review the technical requirements:| Requirement | Detail |
|---|---|
| Min files | 1 |
| Max files | 25 |
| Supported formats | mp3, wav, flac, m4a, ogg, webm |
| Max size per file | 10 MB |
| Recommended duration | 30 – 300 seconds per file |
| Minimum audio quality | 16 kHz, 16-bit |
| Recommended quality | 44.1 kHz, 16-bit or higher |
Clear speech
Clear speech
Recordings must contain natural, clearly spoken speech. Avoid whispering, shouting, or heavy vocal effects.
Single speaker
Single speaker
Each file must contain only one speaker. Multi-speaker recordings will degrade clone quality.
Minimal background noise
Minimal background noise
Record in a quiet environment. If background noise is unavoidable, enable the Remove background noise option before submitting.
Consistent quality
Consistent quality
Use the same microphone and room acoustics across all uploaded files for a coherent clone.
Creating an instant voice clone
Open Instant Voice Clone
Navigate to Voices → Instant Voice Clone (or use the button on the My Voices page).
Enter a name
Provide a name for the cloned voice (max 255 characters). The name is internally prefixed with your account UUID for isolation.
Add an optional description
A short description helps you identify the voice later (max 1,000 characters).
Upload audio files
Select one or more audio files meeting the format and size requirements above. You can upload up to 25 files in a single request.
Optional: remove background noise
Toggle Remove background noise if your recordings have ambient noise. ElevenLabs applies noise reduction before cloning.
After cloning
Once the clone is created you receive avoice_id. In most cases the voice is available for use immediately. Occasionally ElevenLabs may flag a voice as requiring additional verification before it becomes active — the response will include requires_verification: true.
Once created, the cloned voice appears under My Voices and in the voice selector when configuring or editing an agent.
Selecting a voice for an agent
You can choose a voice at two points:- During agent creation — the Configure step includes a Voice field. Click it to open the voice selector modal, browse or search, and pick a voice.
- On an existing agent — open the agent’s configuration page, go to the Voice tab, and choose a different voice. Save the agent to apply the change.
voice_id is stored under conversation_config.tts.voice_id in the ElevenLabs agent payload.
TTS parameters
Alongside the voice itself you can tune the following synthesis parameters:| Parameter | Range | Default | Effect |
|---|---|---|---|
| Stability | 0 – 1 | 0.5 | Higher values make the voice more consistent; lower values add expressiveness. |
| Similarity boost | 0 – 1 | 0.75 | How closely the output matches the original voice sample. |
| Style | 0 – 1 | 0 | Amplifies the speaking style. |
| Speaker boost | Boolean | Enabled | Enhances voice clarity at the cost of slight latency. |