AI voice for museum and tour audio guides
An audio guide lets visitors explore at their own pace: a short track for every room, painting or street corner, often in several languages. Recording it in a studio is expensive, and every new exhibit or corrected date means another session.
SuperClonica voices your audio guide from text, with natural voices, free to test online and offline on your own computer. This guide shows how to write stops that work for the ear, how to voice the whole tour and how to keep it up to date.
Voices for an audio guide
Most catalog voices were recorded in Russian and keep a slight accent in other languages, so test a stop in each language of your guide before you choose.
- The ten stock voices of the model, M1 to M5 and F1 to F5: neutral and free in the engine.
- Conversational Narrator: a neutral male narrator, not too low and not too high, for retellings.
- Anna: a calm, fairly low female voice for letters, memories and personal stories.
- Your curator’s or guide’s own voice, trained from a recording they agreed to, for a guide with a familiar host.
Voice one stop free online
On the page of any voice, paste the first lines of a stop, 30 to 100 characters, choose the language and press Voice it. In a few seconds you can listen and download the recording, with no account.
Write stops for the ear
Read each stop aloud yourself before voicing it: if you run out of breath, the sentence is too long.
- One stop, one idea: one to two minutes is enough for most visitors.
- Start by telling the visitor where to look: “In front of you…”, “On the left wall…”.
- Write names, dates and foreign words the way they should be pronounced, and keep the spelling the same in every stop.
- End with a pointer to the next stop, so the tour flows.
The whole tour offline
Keep every stop in its own text file and install the free SuperClonica engine: python -m pip install superclonica-cpu. Voice each stop with one command, for example superclonica --text-file stop-07.txt --voice F3 --lang en --normalize --output stop-07.mp3. The MP3 files are compact enough to download on a phone or put behind a QR code, and --normalize keeps every track at the same loudness, −18 LUFS. The engine runs offline on an ordinary processor, with no character limit.
Languages, updates and honesty
The same voice reads all 31 languages of Supertonic 3, so every language version of the tour can sound like the same guide; Chinese is not supported yet. When an exhibit moves or a date is corrected, edit one text file and voice that stop again. Tell visitors that the narration is generated, and never put words into the mouth of a real historical person or a living artist with a voice that imitates them.