OpenAI.fm Explained: Voices, Hindi & Urdu, Download
An independent guide by Voicely, updated October 2026
OpenAI.fm is OpenAI's interactive demo for trying the text-to-speech voices in its Speech API. You type a script, pick a voice and a style, and listen. OpenAI's documentation says the speech model handles many languages, including Hindi and Urdu, and that its voices are currently optimized for English. Voicely is a separate service and is not affiliated with OpenAI.
Voicely studio
This is Voicely's own studio, a separate service. It is not the official OpenAI.fm generator and it does not use OpenAI's models. Free to try. Sign in to download generated audio within your account allowance. Studio shows your remaining credits.
What is OpenAI.fm?
OpenAI describes OpenAI.fm as an interactive Speech API demo for trying OpenAI text-to-speech voices. It runs in the browser, and the open-source code OpenAI published for it calls the gpt-4o-mini-tts model. That code is public on GitHub under the MIT license.
Developers who want the same voices in their own app call the Speech API directly, as described in OpenAI's text-to-speech guide.
OpenAI.fm voices and style prompts
OpenAI's text-to-speech guide lists these built-in voices: alloy, ash, ballad, coral, echo, fable, nova, onyx, sage, shimmer, verse, marin and cedar. It recommends marin or cedar for best quality. The older tts-1 and tts-1-hd models support a smaller set.
The open-source demo code lists eleven of these voices, without marin and cedar. We did not check the live site's voice list for this update, so open the demo to see what it offers today.
Style comes from written instructions. The guide says you can prompt the model to control accent, emotional range, intonation, impressions, speed of speech, tone and whispering. In the demo code, each preset “vibe” is a short block of style instructions paired with a voice.
If you publish audio made with OpenAI's voices, OpenAI's usage policies require a clear disclosure that the voice is AI-generated.
Does OpenAI.fm speak Hindi and Urdu?
The API does. OpenAI's guide says the text-to-speech model generally follows Whisper's language support, and that list includes Hindi and Urdu. The same guide says the voices are currently optimized for English.
That is a statement about the API, not a test of the demo. We have not run a dated Hindi or Urdu comparison on OpenAI.fm for this guide, so we make no claim about how its Hindi or Urdu output sounds. The quickest check is your own: paste a line of Devanagari, Nastaliq or Roman Urdu into the demo and listen.
Can you download audio from OpenAI.fm?
The open-source code OpenAI published for the demo includes a Download button next to Play and Share, and handles both WAV and MP3 output. In that code the script and the style prompt are limited to 1,000 characters each.
We could not load the live site when updating this guide (it returned a rate-limit error), so we have not confirmed that the current openai.fm works the same way. If downloading matters for your project, check it on the demo directly.
What OpenAI.fm costs developers
Using the same voices in your own app goes through the OpenAI API, which OpenAI bills. See OpenAI's API pricing for gpt-4o-mini-tts, tts-1 and tts-1-hd.
Need Hindi or Urdu voiceovers for videos? Voicely
If you want an alternative for Hindi or Urdu video narration, Voicely is an independent multilingual text-to-speech studio. It is not affiliated with OpenAI and offers its own browser voiceover workflow, separate from the OpenAI.fm demo.
- • Write in Urdu Nastaliq, Hindi Devanagari, Roman Urdu or Hinglish.
- • No API key: the studio runs in your desktop or mobile browser.
- • audio download. Free accounts have limited downloads. Accounts with a paid-credit balance can download the audio they generate. Available file format depends on the selected model.
- • You can also publish a permanent shareable audio link.
- • Language options vary by Voicely model. Studio shows the options and preview labels for the model you choose.
Free to try. Sign in to use free credits. The Studio shows your remaining credits. Buy credits to keep generating.
Live audio samples
Audio-only preview. Samples rotate on refresh.
Selected voice examples: English, Urdu and Hindi.
All LanguagesOpenAI.fm: frequently asked questions
What is OpenAI.fm?+
OpenAI.fm is OpenAI's interactive demo for trying the text-to-speech voices in its Speech API. OpenAI publishes the code behind it on GitHub under the MIT license. Voicely is a separate service and is not affiliated with OpenAI.
Which voices does OpenAI.fm use?+
OpenAI's text-to-speech guide lists alloy, ash, ballad, coral, echo, fable, nova, onyx, sage, shimmer, verse, marin and cedar. The open-source demo code lists eleven of these, without marin and cedar. We did not check the live demo's voice list for this update.
Can OpenAI.fm speak Hindi or Urdu?+
OpenAI's guide says its text-to-speech model generally follows Whisper's language support, which includes Hindi and Urdu, and that the voices are currently optimized for English. That describes the API. We have not tested the demo's Hindi or Urdu output for this guide.
Can I download audio from OpenAI.fm?+
The open-source demo code includes a Download button and handles both WAV and MP3 output. We could not load the live site when updating this guide, so we have not confirmed that the current openai.fm behaves the same way. Free accounts have limited downloads. Accounts with a paid-credit balance can download the audio they generate. Available file format depends on the selected model.
What does OpenAI.fm cost for developers?+
Using the same voices in your own app goes through the OpenAI API, which OpenAI bills. OpenAI's API pricing page lists gpt-4o-mini-tts, tts-1 and tts-1-hd. This guide does not quote OpenAI's prices.
Is Voicely part of OpenAI or OpenAI.fm?+
No. Voicely is an independent text-to-speech studio. It is not affiliated with OpenAI, with its own browser voiceover workflow rather than the OpenAI.fm demo. This guide links to the official OpenAI.fm demo for anyone who wants the original.
How do I make a downloadable Hindi or Urdu voiceover with Voicely?+
Paste your Hindi (Devanagari or Hinglish) or Urdu (Nastaliq or Roman Urdu) script into the Voicely studio, pick a voice and generate. No API key is needed. Free to try. Free accounts have limited downloads. Accounts with a paid-credit balance can download the audio they generate. Available file format depends on the selected model. The Studio shows your remaining credits.
Compare Voicely vs other AI voice tools
Want a Hindi or Urdu voiceover you can download? Open the Voicely studio and sign in to use free credits.
🎙️ Open Voicely StudioUrdu · Hindi · English · Free to try · audio download after sign-in