cinara

Appearance

Safety and consent ​

Cinara can copy voices and make realistic media. These rules keep that safe. This page explains what Cinara checks and what you agree to.

By using Cinara you agree to use it responsibly, including only cloning voices you have permission to use.

A voice is only cloned from a recording that starts with a one-time consent statement, read by the person whose voice it is.

What you confirm ​

Before you get a statement, you tick: "I am speaker, or I have their permission to clone their voice. I won't use it to impersonate, mislead or harm anyone."

Cinara gives you a statement to read at the start of the recording:

My name is Speaker's full name. I agree to let Cinara create a digital copy of my voice. My one-time code is word word word word.

  • The four code words are chosen at random for each statement.
  • A statement expires 30 minutes after you get it.
  • Each statement works for one recording.

Because the code is new each time, an existing recording of someone can't contain it.

What Cinara checks ​

Before anything is cloned, Cinara listens to the recording. It passes only if:

  • all four code words are heard, in order (small differences in spelling from accents are allowed)
  • at least 80% of the statement is heard
  • only one person is speaking (someone counts as a speaker if they say at least 15% of the words)
  • there are at least 20 seconds of speech

If the check fails, nothing is cloned and your credits are refunded. You see what was heard, for example "Heard 80% of the statement · code not heard · 1 speaker".

Well-known voices are blocked ​

Cinara doesn't clone the voices of well-known people. A speaker name, voice name or description that names a well-known person is refused with "Cinara doesn't clone the voices of well-known people."

The same rule applies across Cinara:

  • Voice Design: "Cinara doesn't design voices that copy well-known people. Describe the voice itself instead."
  • Voice Library renames: "Cinara doesn't name voices after well-known people."
  • Images: "Cinara doesn't make images of well-known people. Describe the scene without naming them."
  • Video: "Cinara doesn't make videos of well-known people. Describe the scene without naming them."
  • Talking Avatars: "Cinara doesn't make talking videos of well-known people."
  • Music styles: "Describe the sound instead of naming a well-known artist."
  • Music lyric ideas: "Cinara writes original songs. Describe the idea without naming a well-known artist."

The name check is a basic guard. The consent statement is what really stops cloning someone without their agreement.

Every cloned voice keeps a record of its consent:

  • the speaker's name, the statement and its code words
  • how much of the statement was heard, how many speakers, and how much speech
  • that you confirmed permission
  • the recording, who made the voice and when

In Voice Library, cloned voices show Consent verified date. Renaming a voice never changes its consent record.

When you delete a cloned voice, the voice and its recording are deleted. The consent details stay on the job record, as an audit trail.

Singing voice recordings ​

A singing voice learns from more recordings of the same person. Each extra recording has its own consent check:

  • Each recording starts with a new consent statement, read by the same person.
  • The statement must name the same speaker the voice was cloned from. Otherwise you see "Get a consent statement for speaker."
  • The same checks apply: the code heard, the statement read clearly, one speaker, at least 20 seconds.
  • If a recording fails the check, it isn't used, and the 100 credits are refunded.

Only use what you have rights to ​

  • Voices: clone only your own voice, or a voice whose owner agreed on the recording.
  • Photos and voices in Talking Avatars: you must tick "I have the right to use this person's photo and voice, and I won't use the video to impersonate, mislead or harm anyone."
  • Songs in Singing Voice: "Use songs you made or have the rights to."
  • Uploads: only upload audio, video and images you own or have permission to use.
  • Dubbing gives each speaker the closest Cinara voice. Nobody's voice is cloned.
  • Images held back by the safety filter aren't shown or charged.

The full rules are in the Terms of Service, the Acceptable Use Policy and the Voice Consent Policy. The Privacy Policy explains how recordings and files are handled.

Share links from Drive are public:

  • Anyone with the link can play the file, or every file in a shared folder, without signing in.
  • Downloads are off unless you tick Allow download.
  • Links can expire after 7 or 30 days, or last until you turn them off.
  • You can see how many views each link has.
  • Choose Turn off to stop a link at once. Removing or deleting a file also stops its links.
  • Treat a link like the file itself. Only share it with people who should have it.

Report misuse ​

Email support@cinara.ai to report:

  • a share link that shares something it shouldn't
  • a voice cloned or used without the person's consent
  • anything else made with Cinara that breaks these rules

Include the link, or the voice's name and where you heard it, and say what's wrong.

Cinara · AI voice, music and video

© 2026 APBLD