Pick the plan that fits you. Sign in or create an account to subscribe.
A monthly credit allowance you spend however you like across TTS, STT, translation and audio intelligence. 1 credit = 1 second of GPU · no trapped quota.
1 The standard voice speaks 9 languages (en, en-gb, es, fr, hi, it, ja, pt, zh), and they are the same ones translation can return audio in. Text is translated into around 50; for the rest, the response arrives as text.
2 The high-quality voice speaks around 30 languages: besides the above, German, Russian, Polish, Dutch, the Nordic languages, Greek, Turkish, Arabic, Korean and several Southeast Asian ones, among others. There is no need to state the language: it is inferred from the text itself.
3 In cloning, the voice sample rules. If it has a strong accent from its own language and you ask it to speak another, the result does not always sound native: it is clearly understood, but you can tell where it came from. It is not a fixed rule — we have heard English samples sound very good in Spanish — it depends on the sample. If the target language matters, test first with a sample in it.
4 Formats accepted across all audio services: wav, mp3, flac, ogg, opus.
Uttera