Voices

April 10, 2024 · Last updated on April 23, 2025

Custom voice cloning best practices

# Voices

-

Creating a natural-sounding voice clone requires more than just a quick voice memo. To ensure the highest quality, follow these smartphone-specific recording tips.

Once you've nailed your voice clone, take it to the next level by creating a custom AI avatar that matches your tone and style. Check out our Best practices for creating your AI avatar guide to complete the experience.

What’s covered in this guide

Audio recording guidelines

Device-Specific Instructions

Common issues and solutions

Additional resources for perfecting your avatar

Audio recording guidelines

To create a natural-sounding voice clone, follow these audio recording tips:

🎙️ Microphone: Use a high-quality external or Bluetooth mic or if you don't have one, a modern smartphone. Avoid using your laptop's built-in microphone. If using an external microphone, it should be positioned about 6-8 inches from your mouth. Ensure the microphone is unobstructed by your hand anddoes not rub against any fabric.

🤫 Environment: Record in a quiet, noise-free space.

🗣️ Speech: Speak naturally and clearly, with slight emotional exaggeration. Discuss familiar topics like daily routines, hobbies, or personal stories. This approach helps you naturally express the tone and energy you’d use in a relaxed setting. If you read from a script, e sure toconvey authentic vocal variation and emotion.

Include breaks and pauses between sentences. Maintain a consistent, clear tone throughout the recording for the best results. Speak at a steady, conversational volume—neither too soft nor too loud. Keep a balanced pace, avoiding rushing or speaking too slowly, and ensure your voice remains controlled and even throughout.

Device-Specific Instructions

iPhone Voice Clone Recording Instructions

This guide helps iPhone users record high-quality voice samples for optimal cloning results.

🎛️ Setup Instructions

Go to Settings → Voice Memos → Audio Quality

Turn on Lossless Recording for the highest fidelity.

🎙️ Start Your Recording

Open the Voice Memos app.

Tap the red Record button 🔴 and follow the script.

Stop and save your recording when done.

✅ Optimization Tips

Speak clearly with exaggerated enunciation.

Keep the microphone 6 inches from your mouth.

Maintain a steady tone—avoid whispering or shouting.

Record in a quiet environment.

⬆️ Upload Instructions

MacBook: Use AirDrop.

Other Devices: Email the file to yourself, then download it to your laptop.

Upload to HeyGen via Browse Local Files.

✅ Supported Formats: .MOV, .MP4, .m4a

🎯 Follow these steps closely for the best results.

For questions, contact: [email protected]

Android Voice Clone Recording Instructions

This guide helps Android users record voice samples with clarity and accuracy for best cloning performance.

🎛️ Setup Instructions

Open the Voice Recorder App.

Navigate to settings and choose High Recording Quality (256kbps, 48kHz).

🎙️ Start Recording

Tap the red Record button 🔴.

Follow the script and stop/save when finished.

✅ Optimization Tips

Speak clearly and pause between sentences.

Hold the device with the mic about 6 inches from your mouth.

Maintain a consistent tone—don’t whisper or shout.

Record in a quiet location.

⬆️ Upload Instructions

Transfer to Laptop:Use email or file transfer (e.g., USB or cloud storage).

Upload to HeyGen via Browse Local Files.

✅ Supported Formats: .MOV, .MP4, .m4a

🎯 Stick to these steps for best results.

For questions, contact: [email protected]

Common issues and solutions

Below are suggestions for how to improve the most common issues our users face in creating voices.

Lip sync

Maintain clear pronunciation and a steady tone. Pause with your lips closed at appropriate moments to improve lip-syncing accuracy.

Video quality

Use high-resolution cameras (smartphones or professional cameras are better than laptop webcams) and ensure proper lighting without harsh shadows. See our Best Practices for Creating your AI Avatar guide for more info.

Too monotone

1. Ensure the audio you upload has natural inflections and expressive tones.

2. First, try cloning your voice separately if you haven't already.

3. If you have, then try recording a new audio sample, sounding very excited and expressive, and add that as a "custom emotion" to your voice.

4. Also try adjusting the stability, clarity/similarity and style exaggeration settings until you like what you hear.

The accent

1. Users who have unique accents and are unable to achieve their desired results using HeyGen's Instant Voice Clone feature should consider creating a Professional Voice Clone (PVC) with a paid ElevenLabs account.

2. Once your PVC is created, integrate it into HeyGen by following these steps. For non-English voices, select the multilingual v2 model in your voice settings in HeyGen. For English, use turbo v2.

1. If issues persist, try re-creating your voice clone with a new sample and choose the "remove background noise" option in HeyGen's voice clone.

2. If issues continue, try regenerating the voice clone a couple of times with the same audio sample.

3. If still unresolved, you can set the accent in the studio. While this fixes the accent, it may make it sound less like your original voice.

Doesn't sound like me

Ensure high-quality, consistent audio with enough training data.

1. If your first voice clone attempt doesn't sound like you, try recording a new audio sample with no background noise and clean audio quality.

2. When you upload the audio in HeyGen, choose the 'remove background noise' option when submitting.

3. If you are still having issues, try regenerating the voice clone with the same sample.

4. If these issues persist, try re-recording another audio sample.

Incorrect pause

1. Adjust your script to simulate pauses (e.g., using punctuation or ellipses).

2. Click the “Add pause section” or use in-text pause by clicking “Add 0.5s pause” to manually add your own pauses.

Pace

1. Modify the script's punctuation or sentence structure to test different pacing effects.

2. Adjust the speed in Voice settings

Incorrect Rise/Fall Pitch

If you’re encountering issues with incorrect rise or fall in pitch, Enterprise users can adjust the punctuation in Proofread to achieve the desired pitch. Modifying punctuation, such as adding commas or question marks, can help refine the intonation.

Recap

By following these guidelines, you can create high-quality avatars with realistic voice and visual fidelity. Have questions or more tips and tricks to share? Join the discussion in this forum post.

This guide covered:

Voice recording guidelines

Device-Specific Instructions

Common issues and solutions

Additional resources for perfecting your avatar

Now that you’ve learned the most important aspects of voice cloning, explore these step-by-step guides to create a personalized avatar that pairs perfectly with your voice:

1. How to use Video Avatars to make your digital twin

2. How to use Motion Avatars to create an animated digital twin

3. How to use HeyGen AI Voice for high-quality video narration

Comments (0)

Popular

Table Of Contents