Resona

Resona is a browser-based voice platform for cloning a voice and turning text into speech. You upload a short sample to build a reusable voice model, or choose from a library of more than 200,000 community voices, then type your script and download the audio. There is nothing to install and no GPU to run.

The site asks for a sample of at least 10 seconds, but its own FAQ calls 30 seconds of clean, single-speaker speech the practical minimum, and says 1 to 3 minutes gives a clearly closer match. Training usually takes seconds to a couple of minutes. Text to speech covers 83 languages, so you can record once in English and generate in Japanese, German or Spanish, though the accent follows your original recording. It also includes emotion tags, WebSocket streaming for voice agents, speech to text with multi-speaker support, and Voice Design, which creates candidate voices from a written description.

There is no subscription. Credit packs start at $9 for 450 credits and never expire, and all features share one balance. A clone costs 100 credits once, a 1,000-word English script about 9 credits, and an hour of transcription about 540 credits. Chinese and Japanese cost more because pricing is by bytes of text. We didn't see a free plan, so you need to buy a pack to generate your own audio.

Consent matters here. Despite the 'clone any voice' headline, Resona's terms ban impersonation and require you to hold the rights or consent for every voice you clone. Only clone your own voice or one you have clear permission to use, and label synthetic speech where the law or platform requires it. The site doesn't name the company behind it, and its terms are governed by Chinese law.

Good fit for course makers and audiobook narrators cloning their own voice, developers building voice agents, and creators who want pay-as-you-go text to speech in many languages.