Voice release
The permission you give before Lamplight records your voice. The app shows this text on its own screen and asks you to agree before recording.
1. Permission to copy your voice
You are about to record your voice. We will use the recording to make a digital copy of your voice (a voice model) and use it only to read stories in your account.
- The recording and voice model are stored by us and by Inworld, which makes the voice model for us.
- We keep them until you delete the voice or your account. If a free account goes 90 days without a story, or a paid account has been cancelled for 12 months, we delete them.
- We never share or sell your voice, and no one outside your account can use it.
- You can delete your voice at any time in Voices. Deleting removes the recording and the voice model from our systems and from Inworld.
By tapping "I agree" you give Empty Bar LLC written permission to collect, store, and use your voice recording and voice model as described.
2. Someone else's voice
On the Pro plan, another adult (a grandparent, for example) can add their voice. The app shows this before their recording:
Hand the phone to the person whose voice this is. They need to read and agree to this themselves, and they will say their name and their permission out loud at the start of the recording. Do not record anyone who has not agreed, and do not record anyone under 18.
The recording must be made in person on your phone. The app does not accept uploaded audio files. A remote recording link is planned for a later version and is not available now.
3. How your agreement is recorded
- Written. The release is on its own screen with its own checkbox. When you agree, the app stores the release version, the name you gave for the voice owner, and the date and time.
- Spoken. The recording starts with you saying: "My name is [your name]. I am recording my voice so this app can make a copy of it to read bedtime stories to my family." This consent line is stored as a separate recording and is not part of the sample used to make the voice copy.
- The consent record (owner name, release version, date) is kept for 3 years after the voice or account is deleted, without any audio, as our proof that permission was given. The spoken consent recording is deleted with the voice.
4. What is made from the recording
- Inworld receives the passage you read (about 30 seconds) and returns a voice model. We store the model's identifier, not the model itself. The model at Inworld is labelled with an opaque identifier, never your name.
- To read a story, we send the story text to Inworld with that identifier and receive audio back. The story text includes the nickname and the names of people, pets, and toys you entered, because the voice has to say them.
- The voice model is used for nothing else. It is never offered to other users, never used in marketing, and never used to say anything you did not generate inside your account.
- What Inworld itself may do with the sample beyond making the model: [confirm on Inworld's current terms and state it here; the spec flags this as a known gap]
5. Where this applies
A voiceprint is a biometric identifier under laws in Illinois, Texas, Washington, and other states. This release is the written notice and consent those laws call for. Our retention schedule is published in the privacy policy, section 5, and deletion is always available on request. [confirm with a lawyer that this release satisfies Illinois BIPA, or decide to block Illinois at signup, per spec decision 9]
6. Withdrawing permission
Delete the voice in the app (Voices, then the voice, then delete), or delete your whole account (Settings, Delete everything), or email support@emptybar.app from your account email. Deleting a voice stops it being used immediately and removes the recording and the model from Inworld and from our storage. Stories already made with that voice stay in your library until you delete them.