HeyGen Voice 1.0

HeyGen Voice 1.0 is HeyGen's text-to-speech model for spoken audio in 39 languages, with the language detected from the text when none is set. It speaks with a public voice or clones a new one from a single recording, and returns the cloned voice so later requests can reuse it. It suits voiceovers, narration for training and product videos, and localized content.

Complete technical specification for integration
Ready-to-use code snippets for common workflows
Step-by-step tutorials for advanced use cases
← All GuidesChoosing a public voice
How to pick one of HeyGen Voice 1.0's 24 public voices for a script, from ad reads and training narration to support lines, wellness audio and announcements.
Introduction
HeyGen Voice 1.0 comes with 24 public voices, each with a name and a style such as Calm & Gentle or Bright & Energetic. You pick one by putting its ID in speech.voice. The choice decides how the text lands: the same shipping notice reads as reassurance in one voice and as good news in another.
Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email.
Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email.
The sections below pair the styles with the content they suit, two candidates per use case, so you can audition a short list instead of all 24. The voice IDs are the same on HeyGen Voice 1.0 Turbo and Turbo Free.
The request
A public voice goes in speech.voice by its ID. The API reference lists every ID with its name and style.
import { createClient } from '@runware/sdk'
const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()
const [result] = await client.run({
model: 'heygen:voice@1.0',
speech: {
text: 'Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email.',
voice: '8be9884ebd16499fbe0efb274e769ed5',
language: 'en'
}
})import asyncio
import os
from runware import Runware
async def main():
async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
results = await client.run({
"model": "heygen:voice@1.0",
"speech": {
"text": "Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email.",
"voice": "8be9884ebd16499fbe0efb274e769ed5",
"language": "en"
}
})
asyncio.run(main())curl https://api.runware.ai/v1 \
-H "Authorization: Bearer $RUNWARE_API_KEY" \
-H "Content-Type: application/json" \
-d '[
{
"taskType": "audioInference",
"taskUUID": "2c8f5a1e-9b37-4d60-a4e2-1f7c3b9d5e84",
"model": "heygen:voice@1.0",
"speech": {
"text": "Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email.",
"voice": "8be9884ebd16499fbe0efb274e769ed5",
"language": "en"
}
}
]'runware run heygen:voice@1.0 \
speech.text="Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email." \
speech.voice=8be9884ebd16499fbe0efb274e769ed5 \
speech.language=en{
"taskType": "audioInference",
"taskUUID": "2c8f5a1e-9b37-4d60-a4e2-1f7c3b9d5e84",
"model": "heygen:voice@1.0",
"speech": {
"text": "Your order has shipped. It should arrive on Thursday, and you can track it any time from the link in your email.",
"voice": "8be9884ebd16499fbe0efb274e769ed5",
"language": "en"
}
}Response
[
{
"taskType": "audioInference",
"taskUUID": "2c8f5a1e-9b37-4d60-a4e2-1f7c3b9d5e84",
"audioUUID": "9a3d6f2b-4e81-4c57-b0a9-3e5f7d1c2b68",
"audioURL": "https://am.runware.ai/audio/os/a14d18/ws/2/ai/9a3d6f2b-4e81-4c57-b0a9-3e5f7d1c2b68.mp3"
}
]Voices by style
The style is the quickest filter. Most styles have more than one voice, so there is usually a second option when the first does not fit the brand.
| Style | Voices |
|---|---|
| Bright & Energetic | Corinne, Ewan, Iona, Jocelyn |
| Upbeat & Lively | Felicity, Marielle, Rory, Sabrina |
| Calm & Gentle | Bram, Nadine, Simon |
| Warm & Friendly | Cecily, Kieran, Weston |
| Conversational & Easygoing | Gareth, Lucian, Tabitha |
| Warm & Soothing | Elowen, Noelle |
| Firm & Measured | Hugh, Soren |
| Clear & Professional | Jonah |
| Serious & Composed | Meredith |
| Thoughtful & Clear | Vincent |
Matching the voice to the content
Product promos and ads
A promo has a few seconds to sell, so it wants a voice that carries energy without help from the script. The Bright & Energetic voices suit sale announcements and short social ads.
This weekend only, every pair of headphones in the store is thirty percent off. Free shipping, a two-year warranty, and delivery in two days. Don't miss it.
This weekend only, every pair of headphones in the store is thirty percent off. Free shipping, a two-year warranty, and delivery in two days. Don't miss it.
Training and e-learning
Instructions have to be followed, not just heard. Clear diction at a steady pace matters more than warmth, which points to Thoughtful & Clear and Clear & Professional for safety training and course modules.
Before you lift any load over twenty kilograms, check the label for its weight, plan your route, and keep the load close to your body as you stand up.
Before you lift any load over twenty kilograms, check the label for its weight, plan your route, and keep the load close to your body as you stand up.
Support lines and onboarding
A phone menu or a welcome screen is often the first voice a customer hears from the company. The Warm & Friendly voices fit IVR greetings and onboarding flows.
Thanks for calling Brightside Internet. To check your current bill, press one. To report an outage in your area, press two. For anything else, stay on the line.
Thanks for calling Brightside Internet. To check your current bill, press one. To report an outage in your area, press two. For anything else, stay on the line.
Wellness and meditation
Breathing cues and guided sessions are played to someone trying to relax, so a slow, soft delivery is the point. Calm & Gentle and Warm & Soothing cover meditation apps and sleep content.
Find a comfortable position and let your shoulders drop. Breathe in slowly through your nose for four counts, then let the breath go for six.
Find a comfortable position and let your shoulders drop. Breathe in slowly through your nose for four counts, then let the breath go for six.
Podcasts and explainers
Long-form audio needs a voice that sounds like talking, not reading. The Conversational & Easygoing voices suit podcast segments and YouTube explainers.
So here's the thing nobody tells you about pricing your first product. The number matters less than you think. What matters is whether you can explain it in one sentence.
So here's the thing nobody tells you about pricing your first product. The number matters less than you think. What matters is whether you can explain it in one sentence.
Announcements and compliance
Building announcements and policy updates have to sound authoritative without alarming anyone. Firm & Measured and Serious & Composed fit public-address messages and compliance training.
Attention, all staff. The north stairwell will be closed for inspection from two until four this afternoon. Please use the east stairwell during that time.
Attention, all staff. The north stairwell will be closed for inspection from two until four this afternoon. Please use the east stairwell during that time.
Tips
-
Audition with the script you will ship. Two candidates reading the real text tell you more than the style labels do, and a line that sounds right in one voice can drag in another.
-
Keep one voice per product. Customers learn a voice the way they learn a logo. Store the ID with the product's settings and reuse it in every request.
-
Split the text for a longer pause. The text is plain, and SSML or
<break>tags are rejected. In our tests periods, commas, ellipses and line breaks all left pauses of under half a second. For a real break, generate the parts separately and add silence when you join them. -
Use the same voice in every language. Each public voice speaks all supported languages. The languages guide covers localizing with one voice.