Text to Speech
Generate natural speech from text and export audio.
Text to Speech
Read any text aloud in natural voices. Export the audio as WAV.
How to use
Tips
Pick the voice language that matches your text for the most natural pronunciation.
Model, languages, and output
Text to Speech uses the Supertonic model in a browser worker. The first synthesis downloads model and selected voice-style assets; keep the tab open and allow time on a slower connection. The language list identifies model modes, but pronunciation quality still varies by language, punctuation, names, abbreviations, numbers, and mixed-language text.
Long text is divided into smaller segments and joined into one result. Add punctuation where a pause is needed, spell unusual names phonetically when appropriate, and preview the whole result before export. Quality and speed controls change processing time as well as sound. The exported file is WAV, not MP3.
Responsible use and troubleshooting
Do not use a generated voice to impersonate a person, mislead listeners, or violate publicity, consent, or intellectual-property rights. For important accessibility, education, medical, or public information, have a person review the script and audio.
If generation stops while loading, check network access to the model host and reload once. For an out-of-memory error, shorten the text, lower quality, close other tabs, or use a device with more memory. If pronunciation is wrong, select the text's primary language and split mixed-language passages into separate exports.
Privacy
Text synthesis runs in a browser worker. The entered text is not sent to an Aiviko processing server; the model, voice data, site code, and provider services can still use the network as documented.