Voice Consent Policy

Voice consent overview

Voice data can capture identity, context, and bystanders in ways text usually does not. This policy sets consent-first expectations for Ethen voice features and should be read with the Privacy Policy, Voice Cloning Policy, and Data Retention policy.

When voice input, recording, or transcription may occur

Voice input may occur when a user intentionally uses a microphone, uploads an audio file, records a spoken prompt, or asks Ethen to transcribe spoken content. Recording or transcription should not be hidden inside a normal text interaction. The user interface should make it clear when voice capture is active, when an audio file is being submitted, and when a transcript may be created. Voice features should be treated as active user choices, not passive background collection.

Notice before recording

Users should provide clear notice before recording or transcribing a speaker. In a one-person workspace, that notice may be obvious from the microphone control. In a meeting, shared room, call, classroom, workplace, family setting, or public environment, notice is more complicated. The person using Ethen is responsible for understanding whether the people being captured know about the recording or transcription and whether additional permission is required by law, contract, workplace policy, school policy, or platform rules.

Other people present

Do not use Ethen to record, transcribe, analyze, or summarize other people without appropriate permission. This includes coworkers, customers, patients, students, family members, contractors, interview subjects, and bystanders. If other people can reasonably be heard in the background, the safest approach is to pause the voice feature or move to a private setting. Voice capture should never be used as a workaround for consent, surveillance, monitoring, or evidence gathering outside an authorized workflow.

Voice data processing

Voice features may process audio, derived transcripts, generation requests, metadata, provider routing information, error states, and usage records. Processing may happen through Ethen systems, configured model or voice providers, local runtimes where supported, or connected services where enabled. Audio may be converted into text, text may be sent to a model, and generated speech may be returned as audio. The exact processing path depends on the feature, account setup, routing mode, provider configuration, and product availability.

Transcription behavior

Transcription can turn spoken content into text that is easier to search, copy, summarize, route, or store. That makes transcription useful, but it also increases privacy risk. A transcript can expose names, credentials, sensitive facts, background speech, confidential business details, health information, financial information, or legally protected communications. Users should review transcripts before sharing them, remove accidental captures where possible, and avoid relying on automated transcription as a perfect record.

Storage, retention, and deletion requests

This policy does not establish exact voice retention timelines. Retention may depend on the feature used, workspace settings, provider behavior, logging configuration, backups, legal requirements, abuse prevention, and final Data Retention Policy decisions. Users may request deletion through available support or account paths once those paths are published. Deletion may not instantly remove every copy from backups, audit records, provider systems, or logs where preservation is legally or operationally required. Final deletion behavior should be governed by the Privacy Policy and Data Retention Policy.

Sensitive voice data

Voice can reveal identity, accent, language, age, mood, health clues, background location, relationships, and emotional state. Users should not submit voice content containing sensitive personal data unless they have authority, a legitimate purpose, and a safe workflow. Regulated workplace, healthcare, legal, financial, educational, government, and security contexts require extra caution. Ethen should not be treated as a substitute for professional review or consent management in those settings.

Workplace, team, and minor use

Teams using voice features should create clear internal rules for recording meetings, transcribing customer calls, using employee voices, saving transcripts, and sharing audio-derived work. Children and minors require stricter handling. Users should not record, transcribe, clone, or generate voice content from minors unless they have the rights, authority, consent, and safeguards required by law and policy. Where uncertainty exists, do not use the voice feature until the consent question is resolved.

Providers, disabling, and related policies

Voice features may depend on third-party providers for transcription, speech generation, storage, routing, or moderation. Provider terms and retention practices may apply separately. Users who do not want voice processing should avoid microphone input, avoid audio upload, remove voice permissions in the browser or operating system where available, and use text input instead. This policy works together with the Privacy Policy, Data Retention Policy, AI Use Policy, Acceptable Use Policy, BYOK Data Handling Statement, and Voice Cloning Policy.

Additional responsibilities for teams

Teams enabling voice features should publish internal rules for meeting recording, customer calls, transcript storage, and sharing. Managers should not require covert recording. Training should cover bystander risk and secure handling of transcripts that may contain credentials or personal data. If a jurisdiction requires two-party consent for recordings, users must comply even if product UI allows capture. Product capability is not legal authorization. Retain only transcripts your organization needs, and delete exploratory recordings promptly.

• Internal rules should cover meetings and calls.
• Local consent laws still apply.
• Minimize transcript retention.

Provider routing for voice

Voice audio and transcripts may be processed by speech or model providers depending on route configuration, including BYOK paths where supported. Provider terms, logs, and retention may apply. Users should avoid dictating secrets into voice prompts when a safer secret entry path exists. See BYOK Data Handling and Privacy Policy for broader routing context.

• Providers may process audio/transcripts.
• Avoid dictating secrets.
• BYOK and privacy policies still apply.

Current voice feature status

Repo code shows hosted speech and transcription endpoints that depend on provider setup, plus voice consent records for cloning gates. This policy stays broader than current feature availability and does not claim every voice feature is live.

• Transcription may require provider setup: The transcription endpoint depends on configured provider credentials.
• Speech generation may use configured providers or mock fallback: Hosted TTS routes depend on configured providers and setup state.
• Consent records are user-declared: The consent gate records declarations but does not independently verify consent.
• Voice cloning is handled separately: Cloning is not available unless explicitly enabled in the product.

Policy status

This is a public-preview voice consent policy. It is consent-first and intentionally conservative.

• Consent first: Users are responsible for notice and permission before capturing other people.
• Feature-bound: Specific behavior depends on released voice features and provider setup.

Not legal advice

These pages are product and policy information for public review. They are not legal advice and should be finalized through qualified legal review before they are treated as binding legal terms.

Proof boundary

This policy does not claim a verified consent process, exact retention timelines, provider retention guarantees, background recording, or live cloning availability.

How to use voice responsibly

Treat every voice interaction as potentially sensitive, especially when other people may be heard.

• Give notice before recording
• Get permission from others present
• Avoid sensitive conversations
• Review voice cloning rules

Frequently asked questions