Personalisation · Audio

My Voices.

Give Sandy a voice you recognise. Record your own voice with explicit consent, preview it and select it for Sandy's future conversations. No voice is cloned without the speaker's own permission.

Checking your Sandy session…

Your voices

Sandy's standard voice is Coral. Custom voices require provider approval. Switching during an active call requires starting a new call.

Create your voice

Checking OpenAI voice access…

Recordings must be of your own voice. The consent statement must be spoken exactly as shown; the sample should be natural speech lasting around 20–30 seconds. Speak in a quiet room.

1. Record your consent

I am the owner of this voice and I consent to OpenAI using this voice to create a synthetic voice model.

Not recorded

2. Record a sample

Read a paragraph or talk naturally about your day. Include a range of ordinary sentences and pauses.

Not recorded

Voice name

This is my own voice. I understand that my recordings will be transmitted to OpenAI to create a synthetic voice model, and that I can revoke its use and request deletion.

Up to 3 active voices per account. Voice provider limits may apply.

Safety: No impersonation or voice cloning of another person through this form. Audio is used only for the chosen provider action and is not added to Sandy's conversation transcripts. Revocation disables the voice immediately in Sandy; provider deletion is tracked separately. Sandy Rooms now requests the selected voice at the start of a new Realtime session, but audible custom-voice delivery has not yet been verified. Other voice-call channels remain on the standard voice.

Please sign in to Sandy and return to My Voices.