AI voice-over for product demos and software tutorials
A product demo goes out of date the moment the interface changes: a button moves, a menu is renamed, and the narration no longer matches the screen. Re-recording a human narrator for every release is slow, and a different voice in the updated part sounds patched.
SuperClonica voices your demo script with the same voice every time, free to test online and offline on your own computer. This guide shows how to split a demo into steps, voice it and keep it current with each release.
Why demos need a voice you can update
Think of the voice-over as part of the product documentation: versioned, reviewed and rebuilt when the product changes.
- Interfaces change every release, and the narration must follow.
- A fixed voice and seed make a re-voiced step sound exactly like the rest.
- The same voice can narrate the demo in up to 31 languages.
- Scripts of unreleased features stay on your own computer.
Test the narrator free online
On the page of any voice, paste the first line of your demo, 30 to 100 characters, choose the language and press Voice it. In a few seconds you hear whether the voice fits your product, with no account.
Write the script step by step
Keep the scripts in the same repository as the product or the docs, so a change in the interface and in the narration go through review together.
- One short text file per step, named in order: step_01, step_02 and so on.
- Say what the viewer does and sees: “Click Export. The report opens in a new tab.”
- Use the exact names from the interface, in quotation marks if they are long.
- Write product names, versions and numbers the way they should be pronounced.
Voice every step offline
Install the free SuperClonica engine: python -m pip install superclonica-cpu. Voice each step with one command, for example superclonica --text-file step_03.txt --voice M3 --lang en --normalize --seed 4 --output step_03.wav. --normalize keeps every step at −18 LUFS, and the fixed --seed means a step voiced again after a release matches the others. Drop the files into your video editor or screen-recording tool and align each one with its part of the recording. In a release script, re-voice only the steps whose text changed.
Choose and keep one voice
Choose a clear, neutral voice: the ten stock voices of the model, M1 to M5 and F1 to F5, are free; Business Male Voice and Young Analyst in the catalog explain calmly without theatrics. For a recognizable brand voice, train one from a recording of a team member who agreed, on a computer with an NVIDIA graphics card, or order it on the exchange. Do not use character or parody voices in product material, and say that the narration is generated where your audience expects to know.