Clone a voice from a short sample.
Create a synthetic version of a voice from a 10-30 second clip, with the speaker's consent, then synthesize any text in that voice by ID.
Included on Starter and above. No training run to wait on.
- 10-30s sample
- Instant, self-serve
- 30+ languages
- Zero-shot and fine-tuned
curl -X POST https://api.speechify.ai/v1/voices \
-H "Authorization: Bearer $SPEECHIFY_API_KEY" \
-F name="Narrator" \
-F gender="female" \
-F sample=@sample.wav \
-F consent='{"fullName":"Jane Doe","email":"jane@example.com"}'Hear a clone before you write a line.
A reference sample in, a cloned narration out. This runs a real clone-then-synthesize request against the API and plays what comes back.
First request
Choose parameters inline, then copy the request.
curl -X POST https://api.speechify.ai/v1/voices \
-H "Authorization: Bearer $SPEECHIFY_API_KEY" \
-F "sample=@reference.wav" \
-F "display_name=Demo narrator"Clone once from a clean 10-30 second sample. Use the returned voice ID in the same speech endpoint you already use for built-in voices.
Voice clone
Reference in, cloned narration out.
Reference voice
Synthesized with cloned voice
Three steps to a cloned voice.
No separate service to integrate. Cloning is an endpoint alongside synthesis on the Build API.
Send a sample
POST a 10-30 second clean clip and a consent record to /v1/voices as multipart form data.
Get a voice_id
The API returns an ID that behaves like any catalog voice, no special handling downstream.
Synthesize anywhere
Pass the voice_id to /v1/audio/speech. One clone speaks 30+ languages via simba-multilingual.
Two tiers, one API.
Start self-serve, or work with us when you need the highest fidelity.
Self-serve cloning
Clone from a single 10-30 second sample via the API or Console. Good quality, ready in moments, no training run to wait on.
Professional cloning
Train on hours of audio for the best possible fidelity. Arranged with our team for signature narrators and brand voices.
Cloned voices run on simba-english and simba-multilingual; the curated simba-3.2 set does not play cloned voices.
Part of Build, on one bill.
Cloning isn't a separate product with separate pricing. Creating and using a cloned voice draws on the same flat per-character Build rate, included on Starter, Pro, and Scale.
Common questions.
What is voice cloning?
It creates a synthetic version of a specific voice from an audio sample, then synthesizes any text in that voice. On the Build API you clone from a short clean sample, with the speaker's consent, and use it by voice ID on the normal speech endpoints, a Build feature that shares the Build API and pricing.
Do I need consent to clone a voice?
Yes. Creating a voice requires a consent record with the speaker's full name and email. It's mandatory: you clone a voice you are authorized to clone, whether it's your own or one you have permission to use.
What languages do cloned voices support?
A cloned voice works across all 30+ supported languages through simba-multilingual, so one clone can speak many languages. Cloned voices also run self-serve on simba-3.0 and simba-english; the curated simba-3.2 voice set plays cloned voices too, but each one currently requires manual Speechify approval.
How do I manage cloned voices?
Retrieve a voice with GET /v1/voices/{voice_id}, download its sample as WAV, and remove it with DELETE /v1/voices/{voice_id}. Cloned voices list before shared voices.
Clone your first voice.
A sample, a consent record, and one API call. Get a key and try it on the Build API.
Cloning included on Starter and above.