Consent and licence terms for synthetic voices

Before you put a synthetic voice in a video or advert, answer two questions. Whose voice was it built from, and does the tool's licence cover how you plan to use the audio? A friendly voice from a menu can still come with conditions. This page can't give legal advice. Ask a qualified adviser or your local regulator. Do it early.

Consent: the person behind the voice

A synthetic voice built from a real person's recordings exists because someone agreed to it. Makers that offer custom voices put the duty on the customer. Microsoft's documentation says you're responsible for your use of the technology and must obtain all necessary permissions, including from voice talent.¹ It adds that customers must get sufficient consent under all relevant laws and regulations from the voice talent.²

Skipping consent has caused trouble in real cases. Wikipedia reports a 2024 one in which a YouTuber noticed their voice was cloned without consent by a company, which later apologized and removed the content.³

A court ruling in Japan, reported in September 2026, found that unauthorized commercial use of an AI imitation of a person's voice can infringe publicity rights.³ That is one country's decision and it doesn't set the rule for the United States, the UK or Australia. Rules on voice and likeness differ by place, so look at your own government's official guidance.

Licence: what the tool lets you do

Even with a stock voice, the maker's terms say where you may use the audio. Look for these points in the terms and the pricing page, and note the date you checked:

• Whether commercial use is allowed on your plan, including adverts and paid training.
• Whether you need to credit the maker.
• Whether the voice or your plan can change or end. ElevenLabs, for example, says its Default voices will expire on December 31, 2026.⁴
• Who owns the finished audio file, and whether you may keep using it after you cancel.

Some voices come from a shared library. ElevenLabs says people can share their voice with the community and set their own terms.⁴ So two voices from the same maker can come with different terms, and you need to read the terms on the voice you pick.

The text you put in

Microsoft also tells customers they're responsible for obtaining any licences, permissions or other rights necessary for the content they input.¹ If you turn a published article, a song lyric or a customer's letter into audio, the voice isn't the only thing that needs permission. Your script needs it too.

Telling listeners

Some makers require you to say that a voice is synthetic. Microsoft requires its customers to disclose the synthetic nature of text to speech voices to its users.² Its documentation also notes that some jurisdictions may mandate this.¹ Whether a law applies to your business depends on your country and your industry, so check the official source for yours.

Keep a record

Save the terms page as they stood the day you hired the voice, the consent paperwork, and the date. If a maker changes its terms later, you'll know what applied when you made the audio. For how makers verify consent for clones, see voice cloning explained.