Let’s explore how you can enhance your videos by adding realistic and expressive Voices. Discover below how Voices can make your content shine.

In HeyGen, a "voice" refers to the AI-generated speech that can be used in your videos, either attached to avatars or as standalone audio. These voices bring your content to life, offering a range of tones, accents, and styles to match the personality of your project.

The Voice Library is your go-to space to manage all your voices in one place. You can review your own voice clones and explore a wide range of public AI voices if you prefer not to use your own.

Our homepage Voice tab has been removed, but you can now access and create voices from three places:

From the AI Studio, when you select a voice for your avatar video, you can now access the Voice Library, listen to previews, and quickly choose the voice that best fits your project.

Interested in a Video Tutorial? Check out our Public Avatars and Voices introduction below.
​

The quality of your voice clone depends directly on the quality of your recording. Here are the most important things to get right:

Go to Settings → Voice Memos → Audio Quality and enable Lossless Recording. This setting is off by default but makes a noticeable difference — compressed Voice Memo files produce lower-fidelity clones.

Open the Voice Recorder app → Settings and select High Recording Quality (256kbps, 48kHz). Android's default recording mode is often insufficient for a high-quality clone.

If you're using an external microphone, position it 6–8 inches from your mouth. Make sure it isn't obstructed by your hand and isn't rubbing against clothing or fabric — fabric noise is a common but invisible cause of poor clone quality.

You can create a Voice cloned directly in the AI Studio.

Just go to the Voice section, click on + New Voice > Create New Voice, and then choose between Instant Voice Cloning or Generate a New Voice.

Interested in a Video Tutorial? Check out our Custom Voice tutorial below.
​

You can still access all of your public and custom voices from the avatar tab as well- just click on any of your avatars and the three dot menu and choose Set Primary Voice and then Manage Voices. This will lead you to your voice library where you can create, delete, or rename new voices.
​

You can also set a primary voice for your avatar, so it would be the default voice anytime you use this specific avatar.

Just click on your Avatar group slot, then on the three dots. You’ll see an option to Set Primary Voice.

Then either select an existing voice, clone a new one, or generate a new voice directly. Once set, that will be the default voice for your avatar!

You can access and manage your voice library from the proofreading as well ( for Team and Enterprise plans only)
Just click on the Track: Voice tab from any of the script boxes. Then you'll be able to review all existing tracks/voices from the current video, as well as the other voices in your library ( you can switch voices on your proofread here!) or create a new voice!

HeyGen offers a range of voice settings to help you find the perfect match for your video. You can filter voices by region, accent, emotion, age, use case, and more — plus, save your favorites for easy access later. These tools make it simple to tailor your avatar’s voice to your audience, content style, and creative goals.

Region: Select your preferred region to hear all the voices most compatible within that selection.

Accent: Select your preferred regional accent.

Filters: Choose from different voice types, use cases, and ages to specify your voice further. You can also filter by your favorited voices [see below].

Favorites: There are a ton of voices in the HeyGen Library; Make sure you mark your favorites so you can filter by them later.

What are Voices in HeyGen, and how can I use them in Scripts?
Voices in HeyGen are AI-generated speech options that you can attach to avatars or use independently within your projects.
​
​What do the different Voice Models Mean?
When using your custom voice on HeyGen, you can choose which ElevenLabs voice model powers it. By default, we use Multilingual v2, a solid all-purpose model that works well across different languages. If you’re looking for faster speaking speed, consider switching to one of the Turbo models—just note that some of them are English-only.

For more information please view this article from ElevenLabs on their different models.

Can I translate my script to be read in another language using one of the Voices?
Yes! You can write your script in any language and seamlessly translate it into another within the Text Editor. Simply click the 'Translate' button in Voice Settings and select your target language.

Can I switch the voice on an existing video?
Yes! this can be done by choosing a different voice track for any / all script boxes inside the proofreading feature ( available for Team + Enterprise plan).

If HeyGen's Instant Voice Clone isn't capturing your accent accurately, the best path is to create a Professional Voice Clone (PVC) through ElevenLabs (requires a paid ElevenLabs account) and connect it to HeyGen.

When setting up the voice in HeyGen, select the model that matches your content:

Selecting the right model is critical — the wrong model can flatten or alter your accent further.

How can I integrate my ElevenLabs [or other third-party] Voice into HeyGen?

HeyGen supports the integration of your ElevenLabs, LMNT, Play, or Character AI Voice, allowing you to enhance your videos further with advanced third-party voice options. To learn how to integrate your third-party voice, check out this article.

HeyGen can generate a unique AI voice from a text description. When writing your prompt, focus on the attributes that matter most:

Start with Age + Accent + Gender for the most distinct results, then layer in optional attributes to refine.

For characters in games, animations, or branded content, simple descriptive prompts work well. Examples:

Other character types that work well: Goblin, Vampire, Elf, Troll, Robot, Pirate, Samurai, Witch, Zombie, Ghost, Knight, and more. Keep the prompt short and evocative — simpler is usually better for character voices.