Acceptable Use Policy

1. Prohibited content and conduct

• Obscene, sexually explicit or sexually demeaning content
• Content that sexualizes minors or is harmful to children and adolescents
• Content promoting violence, abuse or self-harm, or inciting hate or discrimination
• Illegal content that violates the law or promotes or facilitates crime
• Impersonation or imitation of another person’s voice or likeness without consent
• Misinformation, fraud, spam or other content that harms others
• Infringement of copyright, privacy or other rights, or circumvention of access restrictions
• Deceptive audio or impersonation concerning elections, voting, public authorities or emergency notices
• Using or facilitating voice phishing, identity theft, harvesting personal data, non-consensual automated calls or bulk spam
• Delivering individualized medical, legal or financial advice or decisions with significant effects on people without review by a qualified professional
• Submitting, synthesizing or disclosing another person’s confidential material, personal data or sensitive data without lawful authority or a valid processing basis
• Extracting or replicating Company voice models, or supplying or selling actor voices as a reusable voice library or AI training data, without separate permission
• Sharing or exposing API keys, bypassing usage, concurrency or access limits, manipulating credits with multiple accounts or false requests, or overloading the Service

2. Publishing free output and revenue-linked use

Publicly shared free output must carry attribution such as “Voice generated: Soft Story Audio”. Revenue-linked use requires prior consultation and specific permission from the Company for the scope and conditions of use. An inquiry or attribution alone is not permission. No separate agreement permits unlawful use or conduct prohibited by this policy.

• Terms by plan

3. Disclosure of AI-generated audio

If you publish generated audio in a way that could be mistaken for a real person’s or institution’s statement or a recording of a real event, disclose its AI origin in a recognizable way where applicable law requires it. This is separate from free-plan attribution. Do not intentionally remove or alter any AI-generation notice supplied with the output. For fictional or artistic work, use a form of disclosure that preserves the experience where the law allows.

4. API and external integrations

Developers integrating the API into apps, games or hardware must make this policy applicable to end users, provide AI-voice disclosures required by law, and keep API keys out of public clients. They must review abuse reports about their integration and take necessary restrictions. Lawful use of output in a product is permitted, but no right is granted to resell Company voice models or actor voices as a separate voice service.

5. Safety and compliance review

Regardless of plan, the Company may review and process submitted text, generated audio, account information and usage records only as necessary to assess compliance with laws, regulations and guidelines applicable in Korea and service regions, and with the Terms and Acceptable Use Policy; prevent illegal content, rights violations and abuse; and handle reports and disputes. Processing requires a lawful basis under applicable law. Paid users’ text and output remain excluded from training and research to improve speech-synthesis quality. Scope, retention and user rights are governed by the Privacy Policy.

• Data processing and user rights

6. Review and restrictions

The Company reviews reports and verified facts and may request explanation or correction, restrict problematic content or usage, disable an API key, suspend an account or terminate the agreement as appropriate to the case. Urgent action to prevent harm or meet legal duties may occur without prior notice. Reasons and appeal procedures are provided unless restricted by law or investigation or security needs. Paid plans do not exempt users from prohibited-conduct rules, and statutory refund rights remain intact.

7. Reports, consultation and appeals

Contact support to report unlawful use or rights violations, discuss revenue-linked use of free output, or appeal a restriction. Provide only the minimum information needed for review, such as relevant content or its URL, the time and your reasons.

• Contact support

8. AI model protection and prohibited AI-related abuse

The Company retains its rights in voice models, training and inference technology, and service components developed through its own research and development. Third-party materials and open-source components remain subject to their owners’ rights and licenses. Commercial permission for output does not include permission for AI training, distillation, model analysis or research and development. Without the Company’s prior written permission, you must not directly or indirectly use the Service, generated audio, previews or Company-supplied voice materials to train, fine-tune, distill, evaluate, improve or research and develop another AI or machine-learning system, or supply them to third parties for those purposes. This applies to free and paid use, competing and non-competing systems, and commercial, non-profit and academic purposes. The Acceptable Use Policy specifies prohibited conduct and permitted uses.

• Using generated audio or Service responses as teacher-model targets or supervision, or as data or reference values to train, fine-tune, distill, transfer-learn, evaluate or improve AI, including voice cloning, speech synthesis, recognition or speaker identification. This includes extracting and analyzing features, embeddings, labels or transcripts to reproduce or develop models.
• Repeatedly collecting generated audio, previews or voice materials for those AI purposes, or building, publishing, selling or supplying datasets from them, including after segmentation, conversion, augmentation or re-recording. Knowingly commissioning, permitting or assisting third parties to do so is also prohibited.
• Using model extraction, reverse engineering, model inversion or repeated queries to reconstruct or obtain non-public weights, training data, reference audio, speaker features or system prompts, or treating access to the Service as permission to replicate Company or actor voices in a separate reusable model
• Using jailbreaks, prompt injection, evasive inputs or manipulated tool calls to defeat content restrictions or access controls, or obtain secret keys, internal information, or others’ scripts, audio or personal data. Attacks and vulnerability exploitation outside a lawfully authorized security-testing scope are prohibited.
• Rotating accounts, API keys, IP addresses or devices, or using third parties or bots to evade free allowances, concurrency, rate, spending or suspension limits; unauthorized bulk extraction or load testing; or exploiting credit, reward, refund, payment or retry handling to obtain an improper benefit
• Using generated audio to bypass voice authentication or identity checks or impersonate others; deceptively removing or falsifying AI-origin or safety notices; or falsely presenting audio as a real person’s consent, endorsement or actual speech

• AI model and output-use terms

9. Ordinary use and separate permission

These restrictions do not transfer rights in your own scripts or prohibit plan-permitted audio playback in videos, advertisements, games, apps or hardware; ordinary production, editing, captioning or accessibility; checking Service quality; or honest reviews. Inference through ordinary production tools is permitted, but supplying output as training, model-evaluation or research material for those tools or another AI is not. Separate academic or security research or data provision requires the Company’s prior written permission identifying the purpose, materials, target models, third parties and duration. Rights that cannot lawfully be restricted, and independent rights under third-party or open-source licenses, are preserved. Existing prohibitions on unauthorized model replication and rights violations continue to apply; the additional restrictions in this revision follow the amendment notice.