Skip to main content

How to Create a Great Professional Voice Clone (PVC)

Written by Avi

A Professional Voice Clone (PVC) is HeyGen's highest-fidelity voice clone. Instead of a short sample, it learns your voice from 20+ minutes of your own recordings, so it captures your tone, your accent and the way you naturally speak. This guide shows you how to clone your voice with a PVC and get a great result the first time.

💡 Pro Tip: Your clone is only as good as the audio you give it. Clean, natural recordings matter more than anything else.

What do I need to create a Professional Voice Clone?

  • A paid HeyGen subscription plan ( Creator, Pro, Business, etc) plus the PVC add-on ($29/month or $288/year per voice)

  • At least 20 minutes of you speaking. 30 minutes to 2 hours gives the best results.

  • Up to 30 audio or video files in the app (10 through the API), 1.5 GB total, all of you and only you

  • File formats: MP3, WAV or M4A audio. In the app you can also upload MP4 or MOV videos and HeyGen pulls out the audio.

  • Clean audio: a quiet room, no music, no other voices

  • A good mic: an external mic, a Bluetooth mic or a modern smartphone

  • Recordings in one of the 11 supported languages (see the FAQ below), in the language you'll make videos in

How do I record great audio for my voice clone?

Mic

  • Use an external mic, a Bluetooth mic or a modern smartphone. Avoid your laptop's built-in mic.

  • Keep an external mic about 6 to 8 inches from your mouth, and make sure it isn't touching your clothing.

  • Don't set your mic too loud. If your loudest moments distort, that distortion gets baked into your clone.

Room

  • Record in a quiet room. Background noise makes your clone less clear and less consistent.

Delivery

  • Speak naturally and with energy, the way you do in your real videos.

  • Talk about topics you know well, like your work, your routine or a story you love to tell.

  • Reading a script? Vary your pace and use pauses so you don't sound flat.

  • Avoid slow, monotone reading, such as reading lists of words. Your clone will learn to sound flat too.

Length

  • Aim for 30 minutes to 2 hours. The minimum is 20 minutes.

  • Silence counts toward your total, so trim dead air, background noise and any spots where people talk over each other.

Which audio should I use to train my PVC?

Already have hours of yourself on video? That audio can work well, as long as it passes these checks.

  • Match your use case. Train on audio that sounds like the videos you'll make. Making short social videos? Use audio from your short social videos. Making training content? Use a recording of you teaching.

  • Keep one accent and one style. Mixing recordings with different accents or speaking styles can make the accent drift.

  • Include your words. Pick recordings where you say the names, brands and terms you use often, and words that show off your accent.

⚠️ Avoid:

  • Phone-call or video-call recordings. They make your clone sound muffled, like it's on the phone.

  • Music or sound effects under your voice

  • Other people talking, such as interviewers or audience questions

  • Distorted or clipped audio from a mic set too loud

How do I create my Professional Voice Clone?

  1. In HeyGen, go to Avatar > Voices (or start from the home screen).

  2. Click Clone your voice.

  3. Choose Professional Voice Clone ("Our most accurate voice, trained on 20+ minutes.").

  4. Upload your recordings and start training.

Good to know: the PVC flow is upload only, so there is no in-app recording and no consent step. Training usually takes around 15 minutes, but can occasionally take longer.

🛎️ Developers: you can create and retrain PVC voices with the API. See the Professional Voice Clone API docs. Through the API you can upload up to 10 files, capped at 1.5 GB total.

What happens after my voice finishes training?

  • Find your voice. It appears under Avatar > Voices > My voices with a Professional Voice Clone tag. To use it in a video, open the Avatar & Voice panel in the editor and pick it under Voice.

  • Adjust the speed. If your clone reads faster than you'd like, lower the speed. PVC voices have Speed, Volume, Pitch and Inflection controls.

  • Retrain with better audio. Not happy with the result? Delete the voice and create it again with better recordings, using the same slot. Creating it again counts as another training.

  • Know your training allowance. Each voice slot includes five trainings per monthly billing period, pooled across your slots. Your first training counts toward it, and so does every time you delete and recreate a voice. Failed trainings are free.

Why does my voice clone sound off? Troubleshooting

Symptom

Likely cause

Fix

Sounds muffled, like a phone call

Training audio came from a phone or video call

Re-record with a good mic and retrain

Sounds flat or monotone

Training audio was a slow, monotone read, like a list of words

Retrain on natural, energetic speech, like your real videos

Reads too fast

Your training audio is fast, long-form speech

Lower the Speed control

Accent drifts

Training audio mixes different accents or speaking styles

Retrain on recordings with one consistent accent and style

Mispronounces names

Those names aren't in your training audio

Retrain with recordings where you say them

Background noise or clicks

Noise in the room, a mic touching clothing, or a mic set too loud

Record in a quiet room, keep the mic clear of clothing, lower the input level, trim noisy parts, then retrain

Professional Voice Clone FAQ

How much does PVC cost?

$29/month or $288/year per voice, as an add-on to a paid plan.

What's a slot?

Each PVC voice uses one slot. Retraining a voice reuses its slot.

How many voices can I have?

One per slot. To create more professional voice clones, you can purchase more slots.

Can I upgrade my instant voice clone to a PVC?

No. A PVC is created from scratch from your recordings.

Which languages are supported?

PVC supports 11 languages: English, Spanish, Portuguese, French, German, Italian, Russian, Chinese (Mandarin), Cantonese, Japanese and Korean. Other languages may also work, but their quality isn't guaranteed. For the best results, train your voice on recordings in the language you'll make videos in.

Related Articles

Last Updated: October 13, 2026

Did this answer your question?