From script to speech
Open Text to speech in the studio, choose a voice, paste your script, and select Generate speech. Sign in and an active subscription are required before generation. Your estimated credit cost is shown before you generate.
Choose a voice for the job
The studio offers six curated speech voices with short style descriptions to help you choose between warm storytelling, soft guidance, clear narration and conversational delivery. Voices can sound different across languages; choose based on the result that suits your script. All subscription plans use the same voice selection.
Languages and pronunciation
Use automatic language detection or choose English, Chinese, Spanish, French, German, Japanese, Korean, Portuguese, Italian, Russian, Arabic or Hindi. Language selection guides pronunciation; it does not translate the script. Write in the language you want to hear. Accent and pronunciation quality can vary with the voice and text.
Set the pace
Speaking speed ranges from 0.8× to 1.2×. Start at Natural for everyday narration, choose a slower pace for easier listening, or a faster pace for a brisk delivery. Voice, language and speed apply to the next generation. Changing them after generation does not alter an existing audio file.
Understand the cost before generating
Speech uses 1 credit per 80 characters, rounded up for each request. An 81-character script costs 2 credits. Credits are shared with Live Talk. Starter allows up to 2,000 characters per script, Pro 3,500 and Max 5,000. These are per-request limits, separate from the monthly credit allowance. Compare plans and monthly usage.
Download and keep your work
Play the result in your browser and download the MP3. Enable Save this audio before generating to keep a private copy in My library for 30 days. You can rename saved audio and download it before expiry. Saving is optional; unsaved results should be downloaded before leaving the page.
Improve a script before spending more credits
Check names, punctuation and numbers first. Break a longer passage into natural sentences and use a short section to assess pronunciation before generating the full script. Each new generation uses credits; failed generations do not consume credits. The studio currently exports MP3 and does not provide voice cloning or a public developer API.