Skip to main content

Add Custom Voice

A custom voice lets your AI agent speak in a voice you supply rather than one of the built-in variants. Adding one is a review process: you prove you hold the rights to the voice, sign a non-infringement commitment, then submit the materials and a voice file for replication.

Complete these steps in order:

  1. Provide copyright or originality proof of the voice source.
  2. Sign the non-infringement commitment letter.
  3. Submit the materials and voice file for replication.

Choose one of the following methods to prove copyright or originality.

If you hold the copyright yourself, register it in the region where the voice work was created or sold. Because registration takes a long time, you can submit application proof to Tuya first (such as a screenshot of the application website or an acceptance notice) and provide the certificate later.

Method 2: Complete voice IP authorization chain​

If you do not hold the copyright yourself, provide the following:

  • The authorization contract from the original voice copyright owner to you — see the reference attachment. If there is any sub-licensing in between, provide complete authorization proof for each segment.

  • A timestamp certificate. Register and log in to the Intellectual Property Protection Platform, upload the voice file and recording proof, and obtain the original timestamp certificate.

    Add Custom Voice
    Add Custom Voice

Sign the non-infringement commitment letter​

Sign and return the commitment letter.

Submit the materials and voice file​

Self-service upload is not yet supported. Contact your Tuya account manager for help. Before you submit, self-check the following so the materials and files are accurate and consistent:

  • The copyright or originality proof documents are correct.
  • The non-infringement agreement is filled out correctly, with the correct customer signature and seal.
  • The audio preview in the proof documents matches the submitted audio file.
  • You provide the corresponding audio information for voice replication.

Audio file requirements differ by region:

  • China: WAV format, about 20 seconds, 24 K or 48 K sample rate, mono.
  • Overseas: WAV format, under 10 seconds, 24 K or 48 K sample rate, mono. Record the content from this document in the corresponding language.

After you confirm the above, place an order using this value-added service.

See also​