No subscription

Pay-as-you-go AI voice generator

Paste your text and get natural speech back in seconds: voiceovers, podcast intros, phone greetings, narration and lessons, in the language you write. Pay per minute of audio, from $0.0058 a minute, with no plan of minutes to use up.

$0.50 of free credit when you verify your email. Top-ups from $5, and top-up credit never expires.

Coming from a voice plan? ElevenLabs' plans, minutes and rollover, compared with paying per minute

Diese Seite auf Deutsch

Hear it first

Three pieces of text, read aloud

A podcast intro

“Welcome to the Harbor Light podcast. Every week we sit down with the people who keep small coastal towns running: the ferry captains, the bakers who open at four in the morning, the volunteers who repaint the lighthouse every spring. This week we are in a fishing village of three hundred people, where the school has eleven students and the harbor master also runs the post office. We talk about what it takes to stay, why some young people are coming back, and what a town loses when its last shop closes. Pour yourself a coffee, settle in, and let us take you to the edge of the map.”
GPT Audio Mini, voice Nova: 27.5 seconds of speech, ready in 7 seconds. Try this text

A phone greeting

“Thank you for calling Northside Dental. Our office is open Monday to Friday from eight in the morning until six in the evening, and on Saturdays from nine until one. If you would like to book, change or cancel an appointment, please stay on the line and one of our team will be with you shortly. For emergencies outside opening hours, please call the number on our website. We look forward to seeing you.”
GPT Audio Mini, voice Alloy: 20.1 seconds of speech, ready in 5 seconds. Try this text

In German

“Kurz erklärt: Eine Solaranlage auf dem Dach erzeugt tagsüber Strom, den Sie direkt im Haus verbrauchen. Was übrig bleibt, fließt in einen Speicher oder ins Netz, und dafür bekommen Sie eine Vergütung.”
GPT Audio, voice Coral: 14.3 seconds of speech, ready in 5 seconds. Try this text

Made in our own runs on November 10, 2026, through the GenMagic API, exactly as listed: the text as sent, one take each, converted to MP3 for this page.

What $5 makes

One top-up, counted in minutes

39 minutes

GPT Audio

$0.13 a minute of speech at today's price.

How it works

Paste, pick a voice, download

  1. 1. Paste the text

    Exactly what should be said, in the language it should be said in. Commas and full stops shape the pauses.
  2. 2. Pick a voice and a model

    7 voices, the same on every model. The price for your text shows before you press Generate.
  3. 3. Keep it

    The audio is saved to your library at a link of its own, ready to download. Speech that fails is not charged.
Every voice model

What a minute of speech costs

Live prices per minute of audio, cheapest first, and how many minutes a $5 top-up pays for. Need a presenter on screen? A photo and a voice become a talking video on the talking-photo page, and music for the same video is on the pay-as-you-go AI music generator.

ModelPer minuteMinutes for $5
OpenAI: GPT Audio Mini$0.0058854
OpenAI: GPT Audio$0.1339
Questions

About paying per minute

Is there a subscription?

No. You top up a balance from $5 and each piece of speech draws its own price from it, by its length: from $0.0058 a minute on GPT Audio Mini. Top-up credit never expires. There are no monthly minutes to use up and nothing to cancel.

How much does a minute of AI voice cost?

$0.0058 a minute on GPT Audio Mini, $0.13 a minute on GPT Audio, at today's prices. A $5 top-up is about 854 minutes on GPT Audio Mini. The studio shows the price for your text before you generate.

Which voices are there?

7 voices: Alloy, Ash, Coral, Echo, Nova, Sage, Verse, each on every speech model. Pick one by name and it sounds the same in every piece, so a series keeps one narrator.

Which languages does it speak?

The language you write in: the models follow the language of the text. The German sample above was made from German text with no setting changed.

Can I clone my own voice?

No. The voices here are the models' own, and your own voice is not copied. If you need a clone of a specific voice, a voice-cloning tool fits better; for a talking video of you, a photo and a voice you pick work on the talking-photo models.

Can I use the audio commercially?

Yes. You keep the rights you have in what you make, on every account, subject to the terms of the model that made it, with no paid tier needed for commercial use. Where a listener could take the voice for a person's, say that it is AI-generated.

Is there a free trial?

A new account gets $0.50 of credit once its email is verified, about 85 minutes of speech on GPT Audio Mini. It works on every model, with no card needed.

What if it fails?

Speech that fails is not charged. You pay only for audio that comes back.

Can I make speech from n8n, Zapier, my own code or an AI assistant?

Yes, from the same balance: the GenMagic API (POST /api/v1/audio/speech, OpenAI-compatible), a verified n8n node, Zapier and an MCP server for Claude, ChatGPT and Cursor all charge exactly what the studio charges.