​
Overview

PeachDesk Voice Studio can replicate a specific person’s voice from a short audio sample — known as zero-shot voice cloning. You provide 10-30 seconds of clear speech, the model extracts a voice embedding, and from then on you can generate any text in that voice. Everything runs on Bibha’s hosted GPU infrastructure, so there’s nothing to install and no models to download — you simply choose an engine from a dropdown. You create cloned voices in the Voices tab using Clone from audio. Prefer not to record? Use a Built-in voice instead. Every engine in the VAANI family supports cloning, each with different strengths:
Don’t want to record audio? Use a Built-in voice instead — no sample required. See Generating Speech to put a voice to work.

​
How It Works

1

Upload or record a sample

In the Voices tab, choose Clone from audio and provide 10-30 seconds of clear speech from the target voice. You can upload a file or record directly in the browser.
2

Engine analysis

The selected engine, running on PeachDesk’s servers, analyzes vocal characteristics, tone, and speaking patterns.
3

Voice profile created

A voice embedding is generated and stored with your profile, ready to reuse across projects.
4

Generate speech

Switch to the Generation tab, select your new profile, and generate any text in the cloned voice.

​
Choosing an Engine for Cloning

Different engines suit different use cases. Every served engine can clone, so the only thing the engine selector greys out is an engine that doesn’t support your chosen language — switch language or engine and the rest stay available. A voice you cloned on one engine works on the others too: the reference samples are stored with the profile and the voice prompt is rebuilt per engine.
Because every engine is hosted by PeachDesk, you can try a different one at any time without re-downloading or reconfiguring anything — just pick it from the dropdown and regenerate.

​
Best Practices

​
Sample Quality

The quality of your clone depends heavily on the quality of your reference audio.

Do

  • Use 10-30 seconds of audio
  • Clear, consistent speaking
  • Minimal background noise
  • Natural speaking pace

Don't

  • Very short clips (under 5 seconds)
  • Heavy background noise
  • Music or overlapping voices
  • Heavily processed audio

​
Multiple Samples

Adding multiple samples from the same speaker can improve quality:
  • Different speaking styles (casual, formal)
  • Different emotions (happy, serious)
  • Different recording conditions
The model learns a more robust representation from diverse samples. This is especially helpful for distinctive voices the model might otherwise smooth over.

​
Supported Languages by Engine

  • VAANI — English
  • VAANI-ENGLISH — English
  • VAANI-MULTILINGUAL — Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish (23)
  • VAANI-INDIC — Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Tamil, Telugu, plus English (9)

​
Limitations

Voice cloning should only be used with consent. Ensure you have permission to clone someone’s voice, and review your local laws on synthetic voice content. Misuse of cloned voices may violate the PeachDesk terms of service.
  • Quality depends on sample clarity — noisy samples produce noisy clones
  • Works best with a consistent speaking tone within a sample
  • May struggle with extreme accents or speech impediments
  • Background noise reduces quality and can introduce artifacts

​
Next Steps

Generating Speech

Use a profile to generate audio in the Generation tab.

Generation History

Review, compare, and reuse your past takes.
For more, visit docs.peachdeskai.com or open Voice Studio from your PeachDesk dashboard.