HeyGen Voice Alternative AI Text to Speech with ElevenLabs v3
Paste a script, press generate, and download the voiceover. This page uses ElevenLabs v3, a different model from HeyGen Voice.
- 20 preset voices
- up to 10 speakers per script
- 10,000 characters per generation
- 40 credits per 1,000 characters
Quick answer
What is HeyGen Voice?
HeyGen Voice is HeyGen's first in-house text-to-speech model. It launched on October 9, 2026, and runs inside the HeyGen app and the HeyGen API, with no standalone version or third-party licensing. For voice cloning and HeyGen avatar videos, it is the option built into HeyGen's own product. The tool on this page is a different model, ElevenLabs v3, hosted in the Z Image Turbo speech workspace.
| What it is | HeyGen's first in-house text-to-speech model; live on HeyGen and the HeyGen API (HeyGen post, Oct 9 2026). |
|---|---|
| API price | $30 per 1M characters; API users get 50% off ($15/1M) through Oct 31, 2026, applied automatically (HeyGen post). |
| Arena results | Artificial Analysis Controlled Voice TTS Arena: #1 with an Elo of 1,201 across 1,468 arena appearances, ahead of Qwen-Audio-3.1-TTS-Plus (1,182), Eleven v4 Turbo (1,166), and Eleven v4 (1,162). The arena tests the same 8 cloned voices across US and UK English. Pronunciation robustness 83.1% (#10 of 29). Speed: 40 characters per second of generation time (AA post). |
| Voices | Instant clone from one recording (first 3 minutes used, usually ready in seconds); professional clone from 1–10 recordings totalling 20+ minutes, which needs a purchased clone slot. 300+ stock and designed voices in HeyGen's catalog run on the same engine via "Third Party Speech" (HeyGen docs). |
| API details | Completed WAV (44.1 kHz mono) or streamed audio; 1–5,000 characters per request; professional clones support speed 0.75–1.25 and pitch shift ±4 semitones; 30 requests per minute per workspace member (HeyGen docs). |
Sources: HeyGen and Artificial Analysis posts on X (Oct 9, 2026), HeyGen developer docs. Checked Oct 10, 2026.
Side by side
HeyGen Voice vs ElevenLabs v3 on Z Image Turbo
| Feature | HeyGen Voice | ElevenLabs v3 on Z Image Turbo |
|---|---|---|
| Where you use it | HeyGen app and the HeyGen API | Z Image Turbo speech workspace |
| Price | $30 per 1M characters; $15 until Oct 31, 2026 | 40 credits per 1,000 characters, minimum 1 credit |
| Voices | Your clones plus 300+ catalog voices | 20 preset voices, 14 male and 6 female |
| Voice cloning | Yes, instant and professional | No |
| Text per request | 1–5,000 characters | Up to 10,000 characters per generation, 5,000 per line |
| Delivery | WAV file or live stream | MP3 file after generation |
| Account | HeyGen account or API key | Z Image Turbo account; same credits as the image and video tools |
Which one should you pick?
Pick HeyGen Voice when
- You want to clone your own voice, instantly or at professional quality.
- The audio feeds HeyGen avatar videos or other HeyGen workflows.
- You need streamed audio or WAV output through an API.
Pick this tool when
- Preset voices cover the job and you just want to type and generate.
- You write multi-speaker dialogue, up to 10 voices in one script.
- You already use Z Image Turbo and want one credit balance for images, video, and speech.
From script to audio
How to generate speech from text
- 01
Split the script into lines
Each line is one speaker's turn: up to 10 lines, 5,000 characters each, 10,000 total. A single narrator needs just one line.
- 02
Choose a voice and language per line
Assign one of the 20 preset voices to each line. Language stays on Auto detect, or pin one of the 18 supported languages.
- 03
Set stability, then generate
The Stability slider snaps to three levels on the backend: lower is more expressive and variable, the middle is balanced, higher is the most consistent. Generate, preview the audio, and download the MP3.
What people use it for
Text to speech use cases
Video voiceovers
Draft narration for product clips made with the AI video tools on this site.
Explore AI video toolsPodcast and interview drafts
Write a two-host exchange as separate lines, assign voices like Brian and Sarah, and hear the pacing before recording.
E-learning narration
Read lesson scripts in English, Spanish, Japanese, or any of the 18 supported languages, then drop the audio into your course.
Short social ad reads
Generate 15- and 30-second ad voiceovers, compare two voices on the same script, and keep the take that fits.
Before you generate
Limitations
- No voice cloning. The 20 preset voices are the full catalog; for your own voice, use HeyGen Voice's cloning.
- One generation accepts up to 10,000 characters total and 5,000 per line.
- Sign-in is required to generate, and credits are charged per generation.
- Output is an MP3 file. There is no streaming endpoint on this site.
How much does HeyGen Voice cost?
Through the HeyGen API it costs $30 per 1M characters, with a 50% discount ($15 per 1M) applied automatically until October 31, 2026. HeyGen app pricing follows your HeyGen plan. The tool on this page, ElevenLabs v3, costs 40 credits per 1,000 characters.
Can I use HeyGen Voice outside HeyGen?
As of October 10, 2026, HeyGen Voice exists only inside the HeyGen app and the HeyGen API. Check HeyGen's documentation for the latest availability.
Is HeyGen Voice better than ElevenLabs?
In the Artificial Analysis Controlled Voice TTS Arena posted October 9, 2026, HeyGen Voice ranked #1 with an Elo of 1,201, ahead of Eleven v4 Turbo (1,166) and Eleven v4 (1,162). ElevenLabs v3, the model used on this page, was not part of that comparison, so the result does not cover it.
Can I clone my own voice on this page?
No. This workspace offers 20 preset voices and no cloning. For your own voice, HeyGen Voice has instant cloning and professional cloning with a purchased slot.
How many credits does a voiceover cost here?
40 credits per 1,000 characters, with a minimum of 1 credit per generation. A 2,500-character script costs 100 credits; a short 500-character line costs 20.
Which languages are supported?
Auto detect plus 18 languages: English, Mandarin Chinese, Japanese, Korean, German, French, Spanish, Portuguese, Russian, Turkish, Arabic, Hindi, Italian, Dutch, Polish, Swedish, Thai, and Vietnamese.
Do I need to sign in to use this tool?
You can open the workspace and write your script right away. Generating audio requires a Z Image Turbo account, and credits are charged per generation.
Z Image Turbo speech workspace
Turn a script into speech.
Open the speech workspace, paste your lines, pick voices, and hear the result in the browser.