Voices

October 4, 2024 · Last updated on December 10, 2025

Creating and generating custom Voices for lifelike video narration

# Voices

-

Looking for a fast and easy way to create professional voiceovers for your videos?

Meet HeyGen AI Voice, your go-to tool for high-quality video narration. Let’s walk through the simple steps to bring your videos to life with HeyGen’s AI-powered voice engine.

Did you know learners retain up to 80% more information from videos compared to text?

What’s covered in this guide

In this guide, we’ll cover:

What is HeyGen AI Voice?

Voice library

How to create your own voice

Instant voice cloning

Generate AI Voices

How to add a new voice to a video

Prompting guide

Common issues and solutions

What is HeyGen AI Voice?

HeyGen Voice offers a vast library of AI-generated voices that cover a whopping 175+ languages and accents, and capture a variety of emotions that range from friendly, to serious and everything in between. Plus, there are numerous tone, accent, and style options to work with.

Whether you're narrating a corporate presentation, a YouTube video, or an educational course, there's a voice that fits perfectly.

Voice library

Your voice library is where all your voices live in one organized place. Here, you can browse your own voice clones, explore public AI voices, and listen to previews to help you choose the best fit for your video.

When you’re selecting a voice inside the AI Studio, you can simply open the Voice Library, scroll through available options, and instantly apply the one that matches your content. It’s designed to make selecting or switching voices fast and seamless.

How to create your own voice

Start by reviewing our ultimate guide to hyper-realistic Custom Voice Cloning. You can also create a custom voice directly in the AI Studio—even if you’re not making an avatar.

To get started, head to the Voice section and select + New Voice. From here, you can choose:

Instant voice cloning if you want to upload your own audio

Generate a new voice if you prefer to build a brand-new voice with prompts

Both options let you create a personal and expressive voice that fits your style and storytelling needs.

For best results, upload a file that’s 2 minutes long. You can also create an avatar and voice through the Hyper Realistic Avatar creation process.

Finally, click on ‘Create new voice’.

Instant voice cloning

If you want to use your own voice in your videos, Instant Voice Cloning makes it quick and easy to create a personal voice clone.

You’ll see a short agreement confirming that recordings of your voice will be used by HeyGen to create and use a synthetic version of your voice. Once you’ve checked the box, you’re ready to upload your audio.

Click upload file and add the voice recording you’d like HeyGen to use.

After the upload finishes, HeyGen will process your audio and create your custom voice clone.

If the first version of your clone isn’t quite what you had in mind, don’t worry—you can refine it.

Select 'improve this voice' to adjust the accent, voice engine, similarity, style, and stability. These tools help you shape your voice clone so it sounds exactly the way you want.

How to generate AI Voices

You can name your voice, select the desired age, gender, and ethnicity. These details help shape the overall tone and character of the voice.

Write a short description explaining how you want the voice to sound, whether it's professional, friendly, calm, or energetic. If you’re unsure, you can click on "Try a Sample" to hear example voices for inspiration.

Once you’re satisfied with your customizations, click "Generate Voice." HeyGen will provide you with three voice options to choose from. Listen to each and select the one that best matches your needs.

By generating an AI voice, you add a new layer of personalization to your videos, enhancing how your message is delivered!

How to add a new voice to a video

When you’re in the process of creating a video, all you have to do is add the voice to any portion of your script.

Finally, click the portion of the script you’d like to add it to and select the voice you’ve created.

Prompting guide

Voice Design Types

Realistic Voice Design:

To create an original, realistic voice, you can specify attributes such as age, accent/nationality, gender, tone, pitch, intonation, speed, and emotion. Example prompts include:

“A young Indian female with a soft, high voice. Conversational, slow, and calm.”

“An old British male with a raspy, deep voice. Professional, relaxed, and assertive.”

“A middle-aged Australian female with a warm, low voice. Corporate, fast, and happy.”

Character Voice Design:

For creative characters, simpler prompts work well to generate unique voices. Example prompts include:

“A massive evil ogre, troll.”

“A sassy little squeaky mouse.”

“An angry old pirate, shouting.”

Other characters we’ve had success with include Goblin, Vampire, Elf, Troll, Werewolf, Ghost, Alien, Giant, Witch, Wizard, Zombie, Demon, Devil, Pirate, Genie, Ogre, Orc, Knight, Samurai, Banshee, Yeti, Druid, Robot, Elf, Monkey, Monster, and Dracula.

Voice Attributes

Each attribute varies in importance when designing your AI voice:

Age (High Importance): Choose from options like Young, Teenage, Adult, Middle-Aged, Old, etc.

Accent/Nationality (High Importance): Options include British, Indian, Polish, American, and more.

Gender (High Importance): Select from Male, Female, or Gender Neutral.

Tone (Optional): Examples include Gruff, Soft, Warm, and Raspy.

Pitch (Optional): Options like Deep, Low, High, and Squeaky are available.

Intonation (Optional): Options include Conversational, Professional, Corporate, Urban, and Posh.

Speed (Optional): You can set the speed to Fast, Quick, Slow, or Relaxed.

Emotion/Delivery (Optional): Choose emotions such as Angry, Calm, Scared, Happy, Assertive, Whispering, or Shouting.

Common issues and solutions

Doesn’t sound like me

Voice clones may not perfectly replicate the source voice due to issues in the training data or audio quality. To address this:

Use Professional Voice Cloning (PVC) with high-quality, consistent audio.

Record clean audio samples with no background noise.

Enable the Remove Background Noise option when submitting your audio.

If issues persist, re-record new samples or regenerate the clone with the same input.

Pronunciation

Pronunciation accuracy depends on the language of the training data. To improve this:

Record training samples in the language you plan to use.

Ensure natural inflection and clear articulation during recording.

Voice instability

Voice instability can result from inconsistent training data or audio quality. Improve stability by:

Recording high-quality training samples with consistent tone and volume.

Using sufficient training data to provide the AI with varied but stable input.

Incorrect tone

Achieving the desired tone requires tonal consistency in the input samples. To correct tone issues:

Record training samples in the target tone.

Ensure that all samples maintain a consistent emotional or tonal expression.

Too monotone

Monotone voice clones can lack natural expression. Enhance expressiveness by:

Including training data with natural inflections and varied tones.

Recording new samples with expressive delivery and adding them as a custom emotion to your voice.

Adjusting settings like stability, clarity/similarity, and style exaggeration for a more dynamic result.

Emotions don’t sound like me

To capture emotional nuances, training samples must reflect the desired emotional expressions. Include diverse samples showcasing different emotions to enhance the clone's ability to replicate them.

Accent

For unique accents, consistent training data is essential. To address accent issues:

Use Professional Voice Cloning and record samples in the desired accent.

For non-English TTS, use the Multilingual v2 model. For English, use Turbo v2.

If problems persist, regenerate the voice clone using the same or new samples.

Adjust the accent in HeyGen Studio if necessary, keeping in mind this may slightly alter the voice’s authenticity.

Best practices for successful voice cloning

Use high-quality audio: Ensure training samples are free from background noise and recorded in a quiet environment.

Provide sufficient data: Include a diverse range of samples, covering various tones, emotions, and expressions.

Maintain consistency: Ensure all samples are recorded with consistent tone, volume, and quality.

Enable background noise removal: Use HeyGen’s Remove Background Noise option for cleaner input.

Test and adjust: Experiment with settings such as stability, clarity, and style exaggeration to fine-tune the output.

Improving your voice with Voice Doctor

If the voice you created doesn’t quite match the tone or delivery you had in mind, Voice Doctor lets you enhance it without having to rewrite prompts or start over. This tool helps improve clarity, reduce robotic-sounding artifacts, and make your avatar’s voice feel more natural and consistent with the style you intended.

You can open Voice Doctor either from the Studio by clicking on your avatar’s voice, or from your Voice Library by selecting the voice you created and choosing the Enhance Voice option.

Inside the tool, you can describe how you want your voice improved, such as asking for clearer pronunciation, a smoother tone, or more natural pacing, and HeyGen will generate improved options based on your description.

You can also adjust the accent or switch engines if you want a different vocal performance.

Recap

In this guide, you learned:

What is HeyGen AI Voice?

How to create your own voice

How to customize your voice with multiple Emotions

How to add a new voice to a video

Common issues and solutions

We’re looking forward to seeing what you’ll create with HeyGen!

Comments (0)

Popular

Table Of Contents