Skip to main content
Seed Audio 1.0 is live: voices, sound effects and ambience in one take
Voice CloneReference audio with Seed Audio 1.0

AI Voice Clone from a short clip

Upload a clean recording of a voice you have the right to use, and Seed Audio speaks brand-new lines in it. Add up to three voices for a full conversation.

Up to 3 reference voices30-second clipsNo training waitPrivate by default

Voice Clone

Prompt *

0 / 2048

Voice

Auto lets the prompt describe the voice

Reference voices

Up to 3 clips · first 30 s of each

Advanced Options
Speed1.00×
Pitch0 st
Volume1.00×

Format

Sample rate

1 credit per second of finished audio · unused credits return automatically

How to clone a voice

No training run, no waiting room.

1

Upload a clip

Add a clear recording of 5 to 30 seconds with one speaker and little background noise.

2

Write with @Audio1

Refer to the clip in your prompt: @Audio1 says calmly, "Thanks for joining us today."

3

Generate

Seed Audio 1.0 keeps the timbre and speaks the new line. Add @Audio2 and @Audio3 for more voices.

Overview

Voice cloning built into the prompt

Reference clips work like cast members you can direct.

Clone a voice without training

Instant

Clone a voice without training

Seed Audio 1.0 reads your reference clip at generation time, so there is no model training or queue. Upload, tag and generate in the same minute.

Three reference voices in one scene

A full cast

Three reference voices in one scene

Tag @Audio1, @Audio2 and @Audio3 to stage a conversation, an interview or a radio drama with consistent voices, plus the ambience around them.

Your voice, or one you have permission to use

Responsible use

Your voice, or one you have permission to use

Voice cloning is for your own voice, your actors and voices you have consent to use. Clips go straight to the model provider for that take and are not used to train anything.

Clone a voice

Listen

Sample works

Made in the SeedAudio studio. Press play to hear them.

Frequently Asked Questions

Quick answers before you start

1How long should the reference clip be?

Between 5 and 30 seconds of clean speech from one speaker. Longer files are trimmed to their first 30 seconds automatically.

2Can I clone a celebrity voice?

No. Only use your own voice or voices you have permission to use. Impersonating real people without consent breaks our Acceptable Use rules.

3Do you store my voice clip?

Clips are uploaded to our model provider for generation and are not used to train models. Your generated takes stay private unless you share them.

4Can I use more than one voice?

Yes. Upload up to three clips and refer to them as @Audio1, @Audio2 and @Audio3 in the prompt.

5Which file types work?

MP3, WAV, M4A, OGG and most browser-readable audio. The studio converts the clip to clean mono WAV before upload.

6How much does it cost?

New accounts get 60 free credits. Seed Audio takes cost one credit per second of finished audio, and unused credits return automatically. Monthly plans and one-time credit packs are on the pricing page.

Speak new lines in your own voice

Upload a short clip and generate your first cloned take in about a minute.

1