Give every paragraph a voice of its own.

Make spoken audio for product introductions, course lessons, explainers and story narration. Enter the words to be spoken, divide them into paragraphs, and choose the voice and delivery for each paragraph. The studio generates speech, rather than background music.

The studio includes 30 named Google voices, including Kore, Puck, Charon and Sulafat. Built-in previews help you choose a voice character; they are fixed English recordings. The same voice can speak your supported-language script. Custom voices saved in My Voices are also available.

Flash costs 20 credits per 100 characters; Flash-Lite costs 10. Both offer paragraph delivery, built-in and custom voices, and WAV export. Try the same short script and voice with each model, then choose the reading you prefer. The page examples identify which model produced each clip.

Characters in all paragraphs included in one submission are added first, then rounded up to a 100-character block. A 101-character submission costs 40 credits with Flash or 20 with Flash-Lite. Two 50-character paragraphs generated together cost 20 or 10 credits respectively. Generating them separately creates two requests. Regenerating is a new generation; unchanged ready paragraphs are skipped. Whitespace and supported voice-event tags do not count toward billable characters.

You can listen to the fixed built-in voice samples and the examples on this page. To hear your own text and delivery, use the short-test control for up to 120 characters. This sends a model request and shows its credit cost before submission.

Changing a voice updates the next generation, not an existing audio file. Each recording shows the voice used to create that take. Select regenerate with the newly chosen voice; the studio marks changed paragraphs and waits for their new takes before exporting the complete piece.

Keep the spoken words in the script. Choose a delivery preset or put instructions such as “warm, with a measured pace” in the custom delivery field. Insert [pause], [laughs] or other available events at the relevant point in the script. Delivery instructions are sent separately from the spoken words.

Yes. When every non-empty paragraph has an up-to-date recording, export a complete WAV. Choose a 0, 0.3, 0.5 or 1 second gap between paragraphs. You can also play and download individual takes. Editing text, voice or delivery requires a new take for that paragraph before a complete export.

Finished paragraphs stay available. Check the failed paragraph’s error and retry it without remaking the successful paragraphs. The studio submits only missing, failed or changed paragraphs. Failed generations follow the task refund flow; keep the task number if a credit adjustment needs support.

Create or clone a voice in Voice Studio and save it to My Voices, then select it for a paragraph here. Use a recording and voice you have permission to use. For a quick single passage without paragraph controls, use the simpler Text to Speech page.