OpenAI voices and speech model
Choose an OpenAI voice and the OpenAI speech model that speaks it.
Where to find it
Voice › AI voices
Open this screen
Steps
- Open Voice › AI voices and find the card OpenAI voices (the company behind ChatGPT).
- Choose a Speech model. The line under it describes the model.
- Choose a Voice. A voice marked gpt-4o-mini-tts only cannot be used with the other models.
- Press Listen to a sample, then Use this voice.
Settings on this screen
| Setting | What it does | Default |
|---|---|---|
| Speech model | gpt-4o-mini-tts: newest, all 13 voices, follows how-to-speak instructions. tts-1: faster, 9 voices. tts-1-hd: higher audio quality, 9 voices. | gpt-4o-mini-tts unless the deployment sets another |
| Voice | The OpenAI voice. Voices marked recommended are OpenAI's picks for quality. | marin |
Troubleshooting
The card says the OpenAI API key is not set.
These voices can be chosen only after a platform administrator sets the OpenAI API key in the platform console. Choose a voice from another provider meanwhile.
"That speech model cannot use this voice."
Pick a voice the chosen Speech model can speak, or switch the model to gpt-4o-mini-tts.
Tips
- These are the voices of OpenAI's text-to-speech API, not the voice mode of the ChatGPT app.
Good to know
- OpenAI voices take Speed only; Stability and Similarity stay off for them.