vocalove

Clone Voice — Record a Sample, Speak New Words

A short recording becomes new speech in a similar voice. Type the script, generate the audio, and download it. Opens in F5-TTS clone mode. English and Chinese scripts. Add a photo later if you want a talking video.

声音克隆
文案
声音模型
0 / 1000

浏览器里先用欢迎点数试用——无需信用卡。克隆与无水印下载消耗点数；不够时用一次性点数包补充。

What is Clone Voice?

Clone voice turns a short recording of one person into new speech. You supply the sample and the words; the model keeps the timbre and reads the script you write.

This page is Vocalove’s clone-voice workspace. It opens in F5-TTS clone mode: record in the browser or upload a clip, type what should be said, and generate audio you can preview and download. No local install or GPU is required.

A cloned voice is not a stock narrator. Use it when the line should sound like a specific person you have permission to imitate — a greeting, a keepsake, or a draft before a talking-photo video.

  • One short sample

    Three to thirty seconds of clear, single-speaker speech is enough. Record here or upload mp3, wav, m4a, ogg, or aac.

  • Your script

    The sample sets the voice. The text box sets the words. F5-TTS on this page reads English and Chinese.

  • Audio you can download

    Generate returns speech for preview and download. A talking photo is a separate step if you want a face to move with it.

  • Other clone models

    F5-TTS is the model on this page. Qwen3 TTS and Chatterbox Turbo each have their own pages in the Clone Voice menu.

How to clone a voice

The tool at the top of this page is the whole workflow.

  1. Step 1

    Record or upload a sample

    Use Record for a live take, or Upload a file you already have. Aim for 3–30 seconds of one person speaking in a quiet room, with little music or echo.

  2. Step 2

    Write the new line

    Type the words you want heard. Keep the language to English or Chinese for F5-TTS, and confirm you have permission to use the voice before you generate.

  3. Step 3

    Generate and download

    Clone voice runs F5-TTS on your sample and script. Preview the result, then download when it sounds right. Keep the tab open while it runs.

Jump to the clone voice tool ↑

What makes a good voice sample

The model copies what it hears. A clean clip beats a long, noisy one.

  • One speaker only. Overlapping voices or a TV in the background get cloned too.
  • Natural pace. Read a sentence the way the person usually talks; don’t whisper or shout.
  • 3–30 seconds. Shorter than three seconds gives the model too little; longer clips are trimmed toward the speech.
  • Formats: mp3, wav, m4a, ogg, or aac. You can also record straight into the tool.

What a cloned voice is for

Use clone mode when a preset narrator would sound like the wrong person.

  • Personal messages

    Birthday lines, reunion notes, or a short message in a voice people already know.

  • Keepsakes

    Turn a clear home recording into new words of thanks or remembrance, only with permission.

  • Talking photos

    Download the audio, then open Talking Photo Videos if you want a portrait to speak those words.

  • Script drafts

    Hear how a line lands in a similar timbre before you record a final take yourself.

Which model this page uses

The tool above starts on F5-TTS, Vocalove’s zero-shot clone model: one reference clip, no per-person training. Clone mode on this page is about 9 credits per 1,000 characters.

If you want a different clone model, open it from the Clone Voice menu. F5-TTS, Qwen3 TTS, and Chatterbox Turbo are separate pages with their own sample and language notes.

F5-TTS · Qwen3 TTS · Chatterbox Turbo · Pricing

Clone Voice FAQ

What does clone voice do on this page?
It makes new speech from a short sample of one person. Record or upload 3–30 seconds of clear audio, type a script, and generate a downloadable clip in a similar voice. The tool on this page uses F5-TTS.
How long should the voice sample be?
Three to thirty seconds of one person speaking clearly. A quiet room and a steady pace work better than a long clip with music, echo, or other voices.
Which languages can I type?
F5-TTS on this page takes English and Chinese scripts. The reference clip can be either language; write the new line in the language you want spoken.
How much does a clone cost?
F5-TTS clone mode is about 9 credits per 1,000 characters of script. Preset narration on the AI Narration page is a different voice and a different rate. See Pricing for credit packs.
Does clone voice make a talking video?
No. This page returns audio. To move a face with that speech, open Talking Photo Videos and add a portrait you have rights to use.
Whose voice am I allowed to clone?
Only a voice you own or have explicit permission to use. The tool asks you to confirm that before generating. Cloning someone without rights is not allowed.
Can I use a model other than F5-TTS?
This page opens on F5-TTS. Qwen3 TTS and Chatterbox Turbo are listed under Clone Voice in the header, each on its own page.