.hero2__ch{opacity:1!important;animation:none!important}.jf-pending .hero2__lede{opacity:1}
Skip to content
voiced.tw
Trust Center

Privacy Policy

ASWAY LTD
Effective date: 20 October 2026
Download full text (plain text) ↓

Effective date: 20 October 2026

This Policy explains how ASWAY LTD collects, processes, uses, shares, retains and protects data when providing the Voiced website, application and related services, and how you may exercise your rights.

1. Who we are

Voiced is operated by ASWAY LTD, registered in England and Wales under company number 16800650. For personal data whose purposes and means of processing we determine, ASWAY LTD is the responsible company, or “data controller” under data protection law. You may contact us about the use of personal data or your rights as follows.

  • Registered address: Unit A, 82 James Carter Road, Mildenhall, IP28 7DE, United Kingdom.
  • Representative: Ashlyn Ma.
  • Contact and consumer complaints: homepage contact form.
  • Email: support@voiced.tw.
  • UK Information Commissioner's Office (ICO) data-controller registration number: ZC159535.

2. Scope

The voice gateway and registration domain speakbreezily.com are operated by ASWAY LTD and are also covered by this Policy.

This Policy applies to the Voiced website and application, accounts and personalisation, free trials and paid subscriptions, payments, support, feedback and operation of the service. Where a third party independently provides sign-in, payment or another service to you, it may also process data under its own privacy policy; providers processing data on our behalf remain within the disclosures in this Policy.

3. Our data commitments

We do not sell your personal data, provide your data to third parties for advertising purposes, or use your content to train or fine-tune AI models. We require the cloud service providers we engage for speech recognition, text processing and personalization analysis to follow the same restrictions on use.

Audio is released after recognition with zero cloud retention. Content from general text processing and personalization text requests is released once processing is complete. The retention exception for text safety review is handled under Section 5.1.

Local history, personalization settings and learning results that you choose to save, as well as issue reports you actively submit, account data, usage data and billing data, are managed according to the purposes and retention periods in their respective sections.

4. Data we process

The categories of data we process are described below. The listed items are examples; the actual scope depends on the features you use, the data you provide and what is necessary to operate the service.

4.1 Account and sign-in

To create and manage an account, we process account and contact data you provide or that a sign-in service sends, such as your name, email address, profile picture (if provided), verification status, sign-in method, account identifier, and creation or update time. To maintain sign-in and protect the account, we also process authentication and session data, such as tokens, expiry times, IP addresses, and relevant technical information about the browser or application.

4.2 Subscriptions, payments and support

We process subscription plan, price and currency, billing cycle, payment and subscription identifiers, first successful payment time, applicable discounts, payment status, refunds, payment disputes and entitlement-adjustment records. To handle support and billing, we also retain necessary correspondence, support notes, reasons for decisions and administrative action records.

Complete payment-card data is processed directly by the payment provider; Voiced does not store full card numbers or security codes.

4.3 Device, usage and security

To verify usage, deliver plans, maintain quality and prevent abuse, we process device, service-usage and security data, such as a device identifier or its hash, device name, last-used time, request identifier, source IP, version, processing time, word count, audio and text processing volume, status, error category and necessary billing information.

Free trials record periodic and cumulative usage for the account and pseudonymous device to prevent repeat claims. Some requests retain a hash digest of content to identify replay or duplicate processing. A digest is not the original content, but when it can be linked back to an account it is still managed as linkable data. Device signals do not guarantee a person's identity.

4.4 Content you provide

When you use dictation, text-processing and personalisation features, we process the audio, transcription and processed text, instructions, language preferences and context needed to perform those features. For settings, historical text, corrections and related app information used by personalisation features, see Section 4.6.

Content you submit may include personal data about you or others. Please ensure that you have a lawful basis and the necessary permissions to record, submit and use it.

Voiced is a general-purpose voice and text input tool. It is not intended to collect health data or other personal data that receives special protection under the law. Please avoid submitting such data. If sensitive data is incidentally included in general dictation, the transmission and processing of audio and text, and safety review of text content, are handled under Section 5.

We do not use voices to identify a person's biometric identity or use content to build advertising profiles.

4.5 Screen context

“Screen context” must be enabled with your confirmation. You may choose to enable it during setup or later in settings, and may disable it at any time. When enabled, Voiced reads the current window title, focused field and limited excerpts of visible on-screen text during dictation as context for understanding your current input. Those excerpts may contain data about you or others.

We limit context capture using secure-input indicators that the system can recognise and your exclusion settings. Excerpts needed to complete a request are transmitted and processed temporarily under Sections 3 and 5.

Exclusions depend on indicators supplied by the operating system and your settings and cannot recognise every screen that may contain sensitive information. You can disable screen context before processing such content.

4.6 Personalisation settings and learning

Personalisation features process data you provide and settings you enable:

FeatureData used and purpose
Background, vocabulary and style settingsBackground information, specialised vocabulary, common errors, per-app tone and style settings, and custom rewriting instructions you provide, used to choose wording, recognise speech and organise or rewrite text.
Smart personalisationWhen you enable this feature and retain dictation text history, historical text generated through Voiced in different apps and related app information are used to infer domains and commonly used terms and to create or update personalisation settings.
Learn from correctionsWhen you enable this feature, your manual corrections to dictated text and related recognition-error information are analysed to help determine whether specialised vocabulary and common-error settings should be added or updated.

To perform the personalization features you enable, the required settings, historical text, corrections and relevant app information are sent to us and the text processing providers we engage. Relevant text content is released when a general request is complete. Only relevant text content flagged for suspected abuse and requiring further investigation may be retained by text processing providers to the extent necessary for safety review under Section 5.1; this does not include audio.

Local history is retained under Section 6. Personalisation settings and learning results are stored locally under the settings and management features in the application. Voiced does not currently provide cloud storage of this data in your account or cross-device synchronisation.

Custom replacement values are stored and applied locally; only trigger terms needed to complete a request may be transmitted temporarily with that request.

4.7 Performance statistics and diagnostics

We process technical and diagnostic data needed to keep the service secure, troubleshoot it and improve reliability, such as version, latency, processing volume, retries, processing status, performance, error information and necessary cost records. This processing may involve source IP and request metadata listed in Section 4.3. Account-linked records and aggregate statistics that cannot reasonably identify a person are managed separately.

Diagnostic data is managed according to the purposes and legal bases in this Policy and the periods in Section 11. Where consent or an opt-out is legally required for non-essential collection or analysis of device information, we provide the applicable notice and choice.

4.8 Feedback, website and contact

When you actively submit an issue report, we process the original or processed text you choose to send, its description, category, version and submission time. Feedback containing text content is provided only when you choose to enable sharing or actively submit it; it is distinct from technical diagnostics.

We use the content feedback you select to respond, reproduce issues, check errors and verify fixes, and do not use it for model training. Selected content feedback and later quality verification are processed with your consent and retained for the period in Section 11. You may withdraw consent or request deletion as permitted by law.

When you use the contact form, we process your email address, optional name, subject and message to respond, handle complaints or process data-rights requests. When you join a waitlist or request product notifications, we process your email address, interests, language and notification choices. Without separate marketing consent, we send only notices related to what you requested. Website network requests and aggregate statistics are addressed in Section 13.

We use AI tools to help organise support messages and analyse necessary account, payment, error and service-status data, and provide required data to contracted customer-support AI providers. We do not authorise its use for model training. Support data is retained under Section 11.

5. How content is processed and shared

To provide the speech transcription, text organisation or rewriting you request in cloud mode, and the personalisation processing in Section 4.6, necessary audio, text, personalisation settings, related app information and context are encrypted in transit and sent temporarily to our services and contracted speech-recognition, cloud-inference and text-processing providers. Depending on the features you enable, personalisation analysis may separately process retained dictation history or manual corrections. Transmission is limited to what is necessary to perform the relevant feature.

Audio is released after recognition with zero cloud retention. General text and personalization text are released when the request is complete. See the text safety review exception in Section 5.1. Private local mode dictation is not transmitted under this Section unless you actively choose cloud re-recognition for that entry.

Feedback you actively submit, account data and local data are managed under their respective sections. Provider categories and purposes are in Section 9; actual provider names and processing regions are provided on the information page referenced in Section 10.

5.1 Safety review exception

The providers of Voiced’s cloud services for text processing, personalization analysis and related functions may use automated safety classifiers to detect abuse that may violate usage policies, such as assisting phishing, fraud, or creating or distributing malware.

(These are examples and also cover other unlawful or abusive conduct of a similar nature that harms others’ rights or the security of the service.)

When text content is flagged for suspected unlawful or abusive conduct and further investigation is needed, text processing providers may retain the relevant text content to the extent necessary for safety review, and their authorized personnel may review it. Providers state that retention is limited to the period needed for safety review, up to 90 days.

Depending on the content actually sent by each feature, relevant text content may include transcription text, necessary prompts, instructions, settings and context; personalization analysis may also include dictation history text, corrected text and text from relevant app information. This exception does not include audio.

Text processing providers do not use this text content retained for safety review to train or fine-tune AI/ML models.

6. Data on your device

Local history may include transcribed text, processed results, related application information and diagnostic data. Whether local history and audio are retained, the retention period and deletion method depend on your choices during initial setup or later settings. You can adjust the settings or actively delete data using options in the application.

Local data is affected by the security settings of your device and operating system. Voiced does not separately encrypt all local history. You are responsible for device security and backups. Deleting your cloud account does not mean that copies on your device or in backups you created are also deleted.

7. Purposes and legal bases

We process data to provide accounts, dictation, text processing and the personalisation settings you ask us to apply; analyse dictation history or corrections to create or update personalisation settings when you enable smart personalisation or learning from corrections; provide free trials and paid entitlements; handle payments, refunds and support; process content feedback at your choice; maintain reliability, verify usage and prevent abuse; comply with tax and other legal obligations; handle disputes and lawful requests; and send product news at your choice.

Where the UK GDPR or EU GDPR applies:

Processing purposePrimary basis
Accounts, core features, background, vocabulary and style settings you ask us to apply, subscriptions and general supportPerformance of a contract with you or steps at your request before entering into a contract
Necessary security, fraud and abuse prevention, reliability analysis and usage managementLegitimate interests in maintaining a secure and reliable service, balanced against necessity and impact on you; contract also applies where necessary to deliver a plan
Statutory tax, accounting and regulatory requirementsLegal obligation
Establishing, exercising or defending legal claimsApplicable legitimate interests or legal obligation
Optional smart personalisation, learning from corrections, screen context, content feedback and later quality verification, and product marketing messagesYour consent, which may be withdrawn at any time

These bases apply according to the actual data, data subject and processing circumstances; specially protected data also requires the applicable processing condition. Withdrawing consent does not affect lawful processing before withdrawal and does not disable core features that do not rely on that optional data. You may object to processing based on legitimate interests under Section 14.

8. Consequences of not providing data

You may choose not to provide data, but without necessary sign-in information you cannot use account features; without audio or text the related processing cannot be completed; and without information required for payment you cannot subscribe. Smart personalisation, learning from corrections, screen context, text feedback, marketing messages and non-essential personalisation data are optional. Not enabling or providing them does not affect core features that do not need them. History needed for smart personalisation is handled under Section 4.6.

9. Recipients of data

We provide data only to the following categories of recipients where they need it:

Recipient categoryPurpose and scope of data
Cloud service providers that provide features such as speech recognition, text processing and personalization analysis.Process the audio, text, instructions, settings, context and necessary technical data required to perform the features you enable. Personalization features may also involve dictation history, corrections and relevant app information. See Section 5.1 for the safety review exception.
Hosting, database, security and website-operation providersTransmission and storage of account and operational data, and provision of the website, forms, security and aggregate analytics
Payment and identity-verification providersInformation needed for sign-in, payments, invoices, subscriptions and payment security
Communications, support, customer-support AI and operational-notification providersNecessary contact information, correspondence, actively submitted feedback, and account, payment, error and service-status data needed to provide support; ordinary dictation content is not automatically provided for this reason
Professional advisers and authorities legally entitled to receive dataData necessary for tax, accounting, disputes, audits or lawful requests

Contracted providers process data within agreed purposes under applicable service agreements, data-processing terms and our instructions. Where payment or sign-in services process data for their own legal obligations and independent services, they are also responsible under their own policies.

Provider additions, replacements and processing regions are described under Section 10. In a merger, acquisition or business transfer, we provide data only as necessary, require applicable protections to continue, and give notice as required by law.

10. Cross-border processing and provider updates

ASWAY LTD is a UK company. To provide the service, data may be processed by us and our contracted providers in countries or regions outside your location. Principal providers, service purposes, categories of data received, and processing countries or regions are listed on the Sub-processors and processing regions page, which supplements this Policy.

We may add or replace providers for service, security, quality and operational needs. Where the data uses and protection commitments in this Policy do not change, we provide current information by updating that page. We give notice under Section 19 for material changes to data use or protection and obtain consent where required by law.

Where cross-border transfer safeguards are required by law, the applicable safeguards and corresponding service categories are described on the Sub-processors and processing regions page. You may contact us under Section 1 to ask about the safeguards and obtain copies where the law permits. Copies may be redacted as permitted by law to protect others' data or confidential information.

11. Retention and deletion

Data is retained for the periods below and deleted or anonymised when the period expires or the purpose no longer requires it. Where genuinely necessary to comply with a specific legal obligation, resolve an outstanding dispute or address a particular security incident, only directly relevant records are retained longer for the necessary period. This does not permit routine additional retention of ordinary dictation content.

Data categoryRetention method or period
General audioZero cloud retention; released after recognition. Audio you choose to save locally is managed under Section 6 and your settings.
General text and personalization textReleased when a general request is complete. Local data you choose to save and reports you actively submit are managed under the corresponding items in this section.
Relevant text content flagged for suspected abuse and requiring further investigationText processing providers state that relevant text content is retained only for the period needed for safety review, up to 90 days, with no use for model training or fine-tuning. The scope is defined in Section 5.1; audio is not included.
Local history and audioAccording to Section 6 and your settings
Local personalisation settings and learning resultsRetained according to Section 4.6 and the settings and management features in the application
Account dataFor the life of the account; related data is removed from production systems when the account is deleted, while records required by law are handled under the relevant items in this section
Sign-in and security records containing IP addressesGenerally up to 90 days; sessions also become invalid on sign-out, expiry or account deletion
Short-term account-linked performance recordsGenerally 30 days
Technical reliability, request-processing and diagnostic recordsGenerally up to 180 days
Daily account usage and plan-management recordsGenerally up to 730 days; the minimum cumulative and period data needed to enforce free-trial limits are retained for the life of the account
Pseudonymous marker preventing repeat free-trial claims after account deletion24 months from the most recent claim or use of a free trial; the period restarts only when a free trial is claimed or used again
Actively submitted text feedbackWorking and archived copies of original text, processed results and related content for up to 180 days from submission, deleted sooner when no longer needed; long-term test cases use data that cannot reasonably identify a person
General support and contact24 months after the case is closed; feedback content containing original text or processed results remains subject to the 180-day limit above; necessary billing or dispute records follow the applicable obligation or dispute period
AI customer-support working copiesOur copies are deleted within 90 days after the case is closed, and within the same period we request deletion by the provider; provider copies are cleared under applicable data-processing terms and deletion mechanisms. Copies containing text feedback remain subject to the 180-day limit above
Launch waitlistDeleted 90 days after the formal launch notice is sent; you may ask to withdraw or delete it sooner
Separately consented product and marketing notificationsUntil you unsubscribe or consent is no longer valid; only the necessary suppression record is retained to prevent further messages. The waitlist does not automatically become a marketing subscription
Invoices and statutory accounting recordsUsually six years after the end of the relevant accounting year; extended as required by law or for an unresolved investigation, filing or similar matter
Aggregate statistics that cannot reasonably identify a personMay be retained long-term for reliability, quality and capacity trends; not used to contact a person or make individual account decisions

Residual backup copies are used only for necessary recovery, cleared through backup rotation and not used ordinarily. Backup copies of text feedback remain subject to the 180-day limit above.

Usage, security and billing data that can be linked back to an account is not treated as anonymous statistics merely because names are removed, it is encrypted or identifiers are replaced. We do not re-identify data that can no longer reasonably identify a person in order to answer an individual request.

12. Automated restrictions and review

We use payment status, plan terms, usage and security signals to manage entitlements and prevent abuse. An anomalous signal may first trigger request refusal, rate limiting or temporary suspension; it does not mean that a breach has been confirmed.

If you believe a restriction is mistaken, you may object under Section 1. Where the law requires an explanation, human intervention, review or other safeguard for an automated decision, we provide the applicable safeguard.

13. Cookies and website technologies

We use cookies or equivalent technologies needed to maintain sign-in, payments, security and preferences. Aggregate website analytics do not use persistent identifiers to build cross-site advertising profiles, and we do not use advertising pixels to track your dictation content.

Website resources, fonts or embedded services may cause their providers to receive an IP address, browser information, time and resource name. If we introduce non-essential tracking technology that legally requires consent, we obtain consent before enabling it and provide a choice.

14. Your personal-data rights

Under applicable law, you may ask about, access or obtain a copy of your data; request correction, completion, cessation of collection, processing or use, deletion, restriction of processing or data portability; withdraw consent or object to particular processing; challenge qualifying automated decisions; and complain to a supervisory authority.

Submit a request through the methods in Section 1. We may request information necessary to verify your identity. We do not charge where the law requires requests to be free. Where the law permits a necessary fee or refusal of a manifestly unfounded or excessive request, we act under those rules and explain the reason.

Where Taiwan's Personal Data Protection Act applies, we decide a request for inquiry, access or a copy within 15 days of receipt, and a request for correction, cessation or deletion within 30 days. We may extend those periods as permitted by law and give written notice of the reason. Where the UK GDPR or EU GDPR applies, we generally respond within one month, subject to lawful extensions or suspension of time.

These rights remain subject to lawful limits including the rights of others, legal obligations and necessary legal claims.

15. Account deletion

You may request account deletion using the homepage contact form or support@voiced.tw. We process the request after necessary identity verification. Deletion covers account data and related records we retain; statutory accounting records, necessary dispute records and the marker preventing repeat claims under Section 11 are retained for their limited purposes and periods.

When deleting the account, we also stop renewal of the linked subscription and explain when that takes effect. For actively submitted feedback that has already been separated from the account, you may provide a date or necessary clues to help locate it. Local personalisation settings, learning results, history and backups you created are handled under Sections 4.6 and 6.

16. Security

We use measures proportionate to risk, including encryption in transit, storage protection, access restrictions, key management, minimum necessary logging and retention clean-up. Authorised personnel access data only as needed for their work and for limited purposes. No network or storage system can guarantee absolute security.

If a personal-data incident requires notification by law, we notify the supervisory authority and affected people under applicable law and explain the likely impact and response measures.

17. Age

The service is available only to users who meet the age eligibility in Section 2 of the Terms of Service. If we discover an ineligible account, we may restrict or terminate the service and handle related data as required by law.

18. Contact and complaints

If you have questions, a rights request or a complaint about personal-data processing, contact ASWAY LTD through the form, support@voiced.tw or address in Section 1. We handle it under applicable law.

You may also complain to the UK ICO or another personal-data protection authority with jurisdiction.

19. Policy updates and language

This page states the policy version and effective date. For changes that we know about in advance or decide ourselves, and that materially affect your rights or how your data is used, we will notify you in advance through appropriate means, such as the website, the app or email, and obtain consent when required by law.

If an upstream service provider changes arrangements affecting the purposes for which your data is used, how it is retained, or how it is protected without notifying us in advance, we will notify affected users within 3 calendar days after becoming aware of the change, explaining what changed, when it takes effect, its impact and the measures taken. Where the law requires earlier notification or consent, we will follow that requirement.

The latest effective version of this Policy is published on the official Voiced website. If a summary in the app or on another page differs from that version, that version prevails.

This Policy is written in Traditional Chinese. Other-language versions are provided for reference; if there is any ambiguity, the Traditional Chinese version prevails.