Respeak Terms of Use

Version: 2026-11-06.2 · Effective: November 6, 2026 · Operator: xiang zhang

This English version is a translation of the Chinese Terms of Use provided for convenience. If the two versions differ, the Chinese version prevails.

1. Scope and acceptance

These Terms are an agreement between you and xiang zhang ("we", "us") about your use of Respeak (www.getspeakeys.com, the "Service"). By ticking "I have read and agree to the Terms of Use and Privacy Policy" when you sign up or first sign in, you confirm that you have read, understood and agree to these Terms and the Privacy Policy. Please pay particular attention to the clauses in bold that exclude or limit liability, the AI-labelling obligations and the prohibited uses. Minors under 14 may use the Service only with the consent and guidance of a parent or guardian.

2. The Service

The Service lets you fix individual sentences in your recordings: it transcribes the recording, you type the new sentence, and the replacement speech is produced by speech synthesis (including "original voice", which uses the voice of the selected sentence as a reference) or by importing a clip you recorded yourself; the replacement is spliced back into the recording, which you can export as WAV or MP3. Speech recognition and speech synthesis ("original voice") are provided by the third-party provider Alibaba Cloud Model Studio (阿里云百炼). Generated results may contain errors in pronunciation, tone or content, so you should listen to and check each replacement in full before applying it.

3. Accounts

• You must provide true and valid registration information, keep your password and signed-in devices safe, and you are responsible for what happens under your account.
• Different sign-in methods (email, third-party platforms) create separate, independent accounts.
• If your account is compromised or at risk, change your password promptly and contact us.

4. Your content

The recordings, transcripts and content you upload or create belong to you or to the original rights holders. To provide the Service, you authorize us, to the extent necessary, to store, convert, transcribe, synthesize, splice and export this content, and to send it to third-party providers as far as the features require. You warrant that you hold the legal rights to, or have obtained the necessary authorization for, the recordings you upload and the voices and conversations in them, and that they do not infringe anyone's copyright, portrait rights, voice rights, privacy or personal-information rights.

The Service does not keep your recordings for long: recording projects are permanently deleted automatically 3 days after their last edit (7 days for accounts that have bought or redeemed credits), and a project you delete is permanently deleted at once; neither can be restored. Export and keep the results you need in time (exported files stay on the server for at most 2 hours).

5. Voice consent (separate consent)

• "Original voice" generates similar speech using the voice of the selected sentence in a recording as a reference. Original voice may only be used with your own voice; other people's voices (including public figures) cannot be used, even with their consent or authorization. Before using it for the first time on each recording, you must confirm separately in the "Voice consent" window that the voice is your own.
• A consent confirmation is valid for 30 days; you must confirm again when the consent text is updated or the period expires. When you confirm "my own voice" with a personal account, you may choose to have that confirmation cover all recordings you upload to this account for 30 days; if you do, you must make sure that every recording used for original voice during that period really is your own voice, and otherwise change or withdraw the consent in "Voice consent" first. You can withdraw consent at any time; withdrawal does not affect processing completed before it. We keep consent records (including withdrawals) as compliance evidence.
• You must ensure that your confirmation is true, and you bear all consequences of using another person's voice. We cannot verify each confirmation; if we receive a complaint from a rights holder or detect anything abnormal, we may suspend the relevant features, delete content and cooperate with investigations by the authorities.

6. Labelling of AI-generated content

In accordance with China's Measures for Labelling AI-Generated Synthetic Content and the related national standards, the Service labels recordings that contain AI-synthesized segments as follows:

1. Implicit label (file metadata, always added): we write an AIGC label into the exported file's metadata, including the generation attribute (Label), the service provider's name (ContentProducer) and the content ID (ProduceID). MP3 files get an ID3v2 TXXX frame (description "AIGC"); WAV files get the ICMT (comment) field of the RIFF LIST/INFO chunk and a chunk named "AIGC".
2. Explicit label (audio cue): added by default when exporting from the Chinese interface ("Also add the AI audio cue" is ticked) and not added by default in the English interface; you can change it at each export. When added, at the end of the audio we add a "short-long-short-short" cue (Morse code "AI": · — · ·): a 1 kHz sine tone at about −20 dBFS, short tones of 80 ms, long tones of 240 ms and gaps of 80 ms, separated from the speech by 300 ms of silence, about 1 second in total.
3. In the interface: the editor shows "This recording contains AI-synthesized segments" and marks sentences replaced by AI with "AI".

Recordings that contain only clips you imported yourself, and no AI-synthesized content, are exported without these labels.

No audio cue in the Chinese interface: to export without the audio cue from the Chinese interface you must first tick "I understand that the result contains AI-synthesized content, that I will label it as required when publishing it, and that I bear the responsibility for doing so". That version still carries the implicit metadata label. We record your confirmation (account, time, recording, version and export ID) and keep it for at least 6 months.

Your obligations: when you publish audio containing AI-synthesized content online, you must declare that it contains AI-synthesized content as the rules where you are require; where an explicit label is required (for example, content distributed in mainland China that could mislead), add the audio cue when exporting or label it in another way; you must not maliciously remove, alter, forge or conceal the labels above, nor provide tools or help for others to remove them. You are responsible for any consequences of failing to meet these labelling obligations.

7. Prohibited uses

You must not use the Service to do any of the following (see the Acceptable Use Policy, which is part of these Terms); otherwise we may stop the Service immediately, delete the content concerned and report it as required by law:

• impersonate another person, or synthesize or imitate another person's voice (including public figures);
• generate, edit or distribute pornographic, obscene or sexually suggestive (NSFW) content; any sexual content involving minors is strictly prohibited;
• create, alter or distribute content involving another person without their consent, including making them "say" things they never said;
• create harassing, threatening, hateful or discriminatory content, or promote violence, terrorism or self-harm;
• commit fraud, extortion or other crimes, or create or spread false information or rumours;
• infringe others' reputation, privacy, portrait, voice, intellectual property or other legal rights;
• create or spread content prohibited by law, or endanger national security or the public interest;
• remove, alter or conceal AI-generated content labels;
• attack, disrupt or reverse-engineer the Service, bypass access controls or rate limits, or call it automatically at scale.

8. Third-party services and costs

Speech recognition and synthesis always use the third-party speech service we have integrated, at our cost; you do not need to, and cannot, enter your own service keys. While paid features are not enabled, each account has a daily limit on free generations and transcription time, as shown on the page; once paid features are enabled, usage is metered under Section 9 "Paid services (credit packs)". The availability and data handling of third-party services are governed by their own terms.

9. Paid services (credit packs)

This section applies once we enable paid features; until then, what the server speech service can be used for is as shown on the page.

• Credits and credit packs: a "credit" is the unit for using the server speech service (speech synthesis and recognition paid for by us). You can buy credit packs online in the "Credits" window on the website, or in the iOS app; we may also issue redeem codes for promotions or compensation, redeemable in the same window on the website. The number of credits in a pack is as shown on the page; the price, currency and taxes are as shown on the checkout page (for in-app purchases, on the App Store).
• Payment and seller: online purchases are sold by Creem (creem.io, acting as Merchant of Record), which processes the payment, issues the receipt and collects and remits applicable taxes; Creem's buyer terms also apply to the payment between you and Creem. We do not receive your card, payment-account or other payment details. After a successful payment, credits are added automatically, normally within a few minutes if delayed.
• In-app purchases: credit packs bought in the iOS app are paid to Apple; the price is as shown on the App Store, and Apple's terms apply to the payment. We do not receive your Apple ID or payment details. Credits bought in the app are the same as credits bought on the website (used by the same account, with the same validity and rules) and are added automatically after a successful payment. Refunds of in-app purchases are handled by Apple under its policies; request them at reportaproblem.apple.com. After Apple refunds a purchase, we deduct the unused credits of that pack.
• Free credits: each personal account may receive a number of free credits (as shown on the page), valid for 365 days from the day they are granted. The legacy shared workspace does not use credits.
• Validity: credits bought on the website are valid for 2 years (730 days) from the day they arrive, credits bought through Apple in the iOS app never expire, and redeemed credits for 365 days from the day they arrive, in each case as shown by the expiry date on the page; redeem codes must be used within their redemption period. Expired credits and codes lapse and cannot be extended or exchanged for cash. Credits that expire first are used first.
• When credits are charged: 1 credit is charged when you "Replace" with AI speech generated by the server speech service; generations beyond the free allowance are charged as described in the next item. Previews and transcription are free; failed generations or failed replacements are not charged, and failed generations do not use up free generations; importing a replacement clip you recorded yourself is free.
• Free generations: each sentence gets 3 free generations before it is replaced; in addition, each account has a daily limit (China Standard Time) on free generations, as shown on the page. Generations beyond these free allowances cost 1 credit per 3 generations (charged at the first of the three); the count restarts after a confirmed replacement. To control costs, an account needs at least 1 remaining credit to generate with the server speech service; transcription does not need one.
• Usage limits: accounts that have never bought or redeemed a credit pack can upload recordings of up to 10 minutes (30 minutes after buying or redeeming); transcription with the server speech service has a daily time limit, as shown on the page.
• Refunds: to request a refund for a credit pack bought on the website, email lynxzhang1988@gmail.com; we handle it under this section and applicable law and refund through Creem to the original payment method. Refunds are available in one case only: a credit pack none of whose credits has been used can be refunded in full within 14 days of purchase. No other refunds are given, including for a pack of which some credits have been used (its unused credits are not refunded pro rata) and for packs bought more than 14 days ago; credits already used are not refundable or exchangeable for cash, unless the law requires otherwise. If a defect in the Service or a cause on our side keeps you from using credits you bought, you can request a refund for the corresponding unused credits. After a successful refund (including a partial refund) or a chargeback you initiate with your card issuer, we deduct the unused credits of that pack accordingly. Credits from redeem codes are not refundable or exchangeable for cash. Credits charged twice or wrongly through our fault are returned after verification. When you delete your account, any remaining credits lapse with it; to request a refund under the rules above, do so before deleting the account.
• Questions and disputes: for questions about charges, purchases, redemptions or refunds, email lynxzhang1988@gmail.com with your account email, the time it happened and the order number on your Creem or Apple receipt; we normally reply within 15 business days.
• Changes to the Service: we may change credit-pack prices, charging rules or free-credit policies, announcing changes on the page beforehand; changes do not affect the remaining number or expiry date of credits you already have. If we discontinue paid services, we will agree a refund or other reasonable arrangement with you for unused, unexpired purchased credits.

10. Intellectual property

The software, interface design and marks of the Service belong to the operator or the relevant rights holders. Except as expressly permitted by these Terms, you may not copy, modify or use them for other commercial purposes.

11. Disclaimers and limitation of liability

• AI results are generated automatically by models and may be inaccurate or differ from what you expect; you must check them before use and are responsible for how you use them.
• To the extent permitted by law, we are not liable for interruptions or data loss caused by force majeure, third-party service failures, or network or device problems, but we will do our best to limit the loss; please keep copies of important recordings.
• To the maximum extent permitted by law, our liability to you is limited to the fees you have actually paid for the Service (if any).

12. Changes, suspension and termination of the Service

We may change, suspend or end some or all of the Service for operational or legal reasons, giving notice in advance where possible. If you breach these Terms, we may restrict or close your account. You may stop using the Service at any time and delete your personal account yourself in "Account settings" (except the legacy shared workspace); deletion immediately removes the account and all its recordings, any remaining credits lapse, and it cannot be undone. What is deleted and what is kept is described in the Privacy Policy.

13. Changes to these Terms

When these Terms are updated, we update the version number and effective date; for material changes, you will need to read and agree again the next time you use the Service, and you may stop using it if you do not agree.

14. Governing law and disputes

These Terms are governed by the laws of the People's Republic of China (excluding the laws of Hong Kong, Macao and Taiwan). Disputes arising from these Terms should be settled through friendly negotiation; failing that, they shall be submitted to the competent people's court where the operator is located.

15. Contact us

Operator: xiang zhang
Email: lynxzhang1988@gmail.com

If you find someone misusing the Service to impersonate your voice or otherwise infringe your rights, please report it to the email above, with relevant links and evidence where possible.