AI voice over for ads: radio spots, video ads and promos
An ad lives on versions: a 15-second cut and a 30-second one, a new price, another city, another language. Recording each with a voice actor takes time; an AI voice over turns every version into a typed script and a few seconds of synthesis.
SuperClonica has energetic and calm voices with no character behind them, a free online test and a free engine that voices every version on your computer. This guide shows which voices suit ads, how long a script should be, how to voice the versions offline and what to keep in mind before an ad goes on air.
Voices for ads
Most catalog voices were recorded in Russian and keep a slight accent in other languages, so voice the key line of your ad in your language before you choose.
- Energetic Katya: a pleasant, energetic young female voice.
- Energetic Narrator: mid-low with clear emphasis, so a promo sounds alive rather than official. No specific host behind the model.
- Young Analyst: reads comparisons and product benefits clearly, without theatrics.
- Pleasant Announcer: low and restrained, for a calm brand voice.
- The ten stock voices of the model, M1 to M5 and F1 to F5: free in the engine.
Test a line of your ad free
On the page of any voice, paste the key line of your ad, 30 to 100 characters, choose the language and press Voice it. In a few seconds you can hear whether the voice sells it and download the recording. No account is needed.
Write a script that fits the time
Time every version after voicing it: a few words too many is easier to cut in the script than in the audio.
- About 2 to 2.5 words a second: roughly 30 to 35 words for 15 seconds and 65 to 75 for 30.
- One message per spot, and the brand name early and again at the end.
- Prices and numbers as you want them read; the engine spells digits out in 27 languages, so check the result by ear.
- A clear call to action in the last sentence.
Every version offline
Install the free SuperClonica engine: python -m pip install superclonica-cpu. Keep each version of the script in its own text file and voice it with one command: superclonica --text-file spot-15s.txt --voice M3 --lang en --normalize --output spot-15s.wav. The --normalize option brings all versions to the same loudness, −18 LUFS, and the result is a 16-bit WAV for any editor. It runs offline on an ordinary processor, with no character limit, so a new price or a new city costs nothing extra.
Before the ad goes on air
Do not use character voices or voices that imitate real people in ads: a fan-made or parody voice is not an endorsement. For a lasting brand voice, record a clean phrase of 5 to 10 seconds of a speaker who agreed in writing, train it in the studio on a computer with an NVIDIA graphics card, or order it on the exchange. Some platforms and countries require AI-generated content to be labeled, so check the rules where the ad will run.