Artlist AI Voiceover and Voice Cloning: Ten Takes You Can Hear (2026)
Last verified: 2026-11-08
Voiceover is the cheapest of the AI tools in how Artlist works: an 8-second line cost 11 credits on one model and 22 on another, same voice, on our Max Pro account.
How do the voice models compare?
Same voice ("Suburb"), same line, English, American accent, no effect:
Every morning, the first light reaches the pond before anything else wakes up. Then, one by one, the birds begin.
| Model | Rate on screen | Credits for this line | File |
|---|---|---|---|
| Eleven v3 | 1 credit = 5 characters | 22 | 8.0 sec MP3, 128 kbps |
| Eleven v4 | 1 credit = 8 characters | 14 | 8.0 sec MP3, 128 kbps |
| MiniMax 02 HD | 1 credit = 10 characters | 11 | 8.4 sec MP3, 259 kbps |
Eleven v3
Eleven v4
MiniMax 02 HD
Artlist's model list also carries Eleven Multilingual v2 (1 credit = 5 characters) and Cartesia Sonic 2 (1 credit = 10 characters).
For how these rates fit next to images and video, see Artlist AI credits.
How do you make a voiceover?
- Open Voiceover in the left menu of the AI tools.
- Type your text. The AI Voiceover FAQ allows up to 5000 characters per generation.
- Pick a model and a voice.
- Open the settings for language, accent, stability and effect.
- Hold the pointer over Generate for the price, then press it.
Which voices can you pick?
The Voice Catalog has three tabs: Artlist Original Voices, External Voices and Custom Voices. You filter by model, language, category, gender and age, and each voice carries tags such as Documentaries, Tutorials or Trailers.
What do stability and effects do?
Stability's tooltip reads: "Lower values produce more expressive output, higher values produce more stable output." It sat at 50% by default on Eleven v3. The Effect list on that model opened on No Effect, followed by nine others from Vintage Radio to Walkie-Talkie, each with a play button.
The FAQ adds that Eleven v3 reads audio tags: write any text in brackets before the words it should direct, for example "[Calm] Good morning!".
Same script, three voices
A longer line, 170 characters, the length of a short ad read:
Our bakery opens at six. The first loaves come out of the oven at five, and by seven the shelves are full. Come early on Saturdays: the sourdough is usually gone by noon.
We ran it on Eleven v4 with three voices from the catalog, each on its default settings. Each cost 22 credits.
| Voice | Accent it picked | Length |
|---|---|---|
| Suburb | American | 10.8 sec |
| Path | American | 10.9 sec |
| Professor | British | 11.7 sec |
Suburb
Path
Professor
The accent setting follows the voice: Professor came back British without our asking. Set it yourself if your other lines are in a different accent.
What does stability change in practice?
We ran the same script on Eleven v3 with Suburb twice, stability at 0 and at 100%. Both cost 34 credits.
| Stability | Length | Pauses longer than 0.3 sec |
|---|---|---|
| 0 | 11.7 sec | 1 |
| 100% | 13.1 sec | 3 |
At full stability the read was 1.4 seconds longer, with three pauses over 0.3 seconds against one at 0.
Stability 0
Stability 100%
Do audio tags cost extra?
They count as characters. We added [whispers], [excited] and [laughs] to the same line on Eleven v3, which took it from 170 to 200 characters and the price from 34 to 40 credits. The take ran 13.0 seconds.
[whispers] Our bakery opens at six. [excited] The first loaves come out of the oven at five, and by seven the shelves are full! [laughs] Come early on Saturdays: the sourdough is usually gone by noon.
Do effects cost extra?
No. Path with Vintage Radio on Eleven v4 cost 22 credits, the same as without. The file differed from the plain take: 48 kHz instead of 44.1 kHz, and 7 dB less energy above 5 kHz. It was also a new read, 11.7 seconds against 10.9, not the same take with a filter on top.
What does the same line cost on each model?
The 170-character script, priced without generating:
| Model | Credits |
|---|---|
| MiniMax 02 HD | 17 |
| Eleven v4 | 22 |
| Eleven v3 | 34 |
| Eleven Multilingual v2 | 34 |
Eleven v4's label says 1 credit = 8 characters, which would make this 21. We were quoted 22, and 133 for a 1,025-character script where the label gives 128. Budget a few percent over the label on v4.
How does Artlist voice cloning work?
Custom Voices starts empty, with this promise: "Turn any voice recording into a professional AI voice from just 10 seconds of audio".
Create Custom Voice opens an upload box. It takes 10 seconds to 5 minutes of MP3, WAV or M4A, up to 20 MB, and says "You can record in any language".
Through MCP, creating a custom voice costs 500 credits. The FAQ says voices one team member uploads are visible and usable by the entire team. For generating with a clone, the tooltip on this screen lists languages including English, Mandarin Chinese, Cantonese, Arabic, Hindi and Ukrainian.
What comes after the voiceover?
The finished line can drive a talking clip: see Artlist lip sync, avatars and dubbing. An Artlist discount code gives 2 free months on annual plans. Weighing the other big music library? Here is the Epidemic Sound discount code.