1. Who’s Responsible for Your Data

1. Who’s Responsible for Your Data

SpeechPath and SpeechPath Lab are built and run by Myung Joon Kang, an independent student developer based in Korea. We’re the one responsible for the data described in this Policy. There’s no company or team behind it.

Privacy officer: Myung Joon Kang, privacy@myungjoon.com.

2. What SpeechPath Lab Is

2. What SpeechPath Lab Is

SpeechPath Lab is a way to help build SpeechPath, starting with SpeechPath 2.0, a version of SpeechPath that’s in development and hasn’t been released. You say short phrases while your iPhone’s front camera tracks your lips, and you can send each recording of a phrase (a “take”) to help build SpeechPath’s lip guides. The SpeechPath available today doesn’t use this data.

3. What We Collect

3. What We Collect

When you send a take, it contains:

Lip and face tracking from Apple’s ARKit, frame by frame: the outline of your lips, the mouth’s blend shapes (how open, rounded or stretched it is), and a 3D mesh of your whole face’s shape. Lip and face tracking may count as biometric data under some laws, including in Illinois, Texas, Washington, the EU and Korea. We never use it to identify or recognise anyone.

Sound level, only when Sound Check is on. This is a loudness reading, never a recording.

The phrase: its ID and text, and how many frames the take has.

Your native language and accent, as you chose them.

Context: the consent version you agreed to, a random ID that groups one sitting’s takes, the take’s number in that sitting, the date and time it was recorded, your iPhone model (such as iPhone16,1) and SpeechPath Lab’s version.

Review status: whether the take is waiting for review, used or rejected.

An iCloud ID that Apple’s CloudKit adds to every record. iCloud gives your account this ID for SpeechPath Lab only, so other apps get a different one, and it doesn’t reveal your name, email or Apple Account. It’s what lets you see and delete only your own takes. SpeechPath Lab shows it to you as your Contributor ID, so you can quote it if you ask us to delete your takes. A different Apple Account has a different ID, so takes you sent before keep the ID printed in your consent PDF from then. Please keep it private.

Your age check, language, accent and consent are also kept on your iPhone so SpeechPath Lab can remember them. For your consent PDF only, it also keeps on your iPhone how the age range Apple shared was set: by you, by a guardian, or confirmed by Apple. When you agree, you sign with your finger. Your signature is kept only on your iPhone, as the pen strokes, with the date and version you agreed to. It’s never uploaded, so we never receive it. Withdrawing consent or leaving SpeechPath Lab deletes it, and Leave SpeechPath Lab clears the rest too. You can export a PDF of your consent and signature from Settings, which you share yourself. If you use VoiceOver, you can agree without signing. When you agree, a record of it is also saved in SpeechPath Lab’s database: the consent version, the date, whether you signed or agreed with VoiceOver, SpeechPath Lab’s version and your language setting. It never includes your signature and, like your takes, carries only your iCloud ID.

4. What We Don’t Collect

4. What We Don’t Collect

No video or photos. The camera’s image isn’t saved or sent.

No audio. Sound Check reads a sound level only.

No name, email address, phone number or Apple Account details.

No location, contacts or advertising identifiers.

No tracking across other apps or websites.

5. How We Check Your Age

5. How We Check Your Age

SpeechPath Lab is for adults only. Through Declared Age Range, Apple shares the age range on your Apple Account, never your birth date, and recording stays off unless that range is 18 or over. SpeechPath Lab keeps the answer on your iPhone. It isn’t sent with your takes.

6. How We Use It

6. How We Use It

Every take is reviewed before it’s used. Takes that can’t be used are marked rejected, and SpeechPath Lab shows them to you as Not used. A later review may still use them.

Takes that pass review are used only to build, test and improve lip guides for SpeechPath, starting with SpeechPath 2.0, which is in development and not yet released.

Phrase, language and accent are used to tell lip guides for different languages and accents apart.

We never sell your data, use it for advertising, build a profile of you, make automated decisions about you, or use your face data to recognise or identify anyone. A lip guide is a new visualisation built from takes, never an exact copy of anyone’s take.

7. Why We’re Allowed to Use It

7. Why We’re Allowed to Use It

We rely on your consent. Before you can record, SpeechPath Lab shows what’s saved and asks you to agree by signing, and each take records which version you agreed to. You agree separately, with a switch that’s off until you turn it on, to SpeechPath Lab saving the shape of your face, which may count as sensitive or biometric data. Under the GDPR this is consent under Article 6(1)(a) and, because face data may be biometric, explicit consent under Article 9(2)(a). You can withdraw it at any time without any disadvantage, and withdrawing doesn’t affect what was done before.

8. Where It’s Stored

8. Where It’s Stored

Sent takes and consent records are stored in Apple’s iCloud, in a CloudKit public database that belongs to SpeechPath Lab. Apple hosts it as a service provider. Its access rules let you read, change and delete only your own takes and consent records, and only the developer’s role can read everyone’s. If you use the request form, what you send is stored by Framer B.V. in Amsterdam, which hosts this website and processes it for us under its Data Processing Addendum. Framer’s hosting providers may store it outside the EU.

To review takes, we download copies and keep them on our own encrypted computer, where they’re analysed. Only combined results, such as averages and counts, may be shared with Claude, an AI assistant made by Anthropic, to help build lip guides. Combined results are built only over fixed groups of at least 10 contributors, without minimums, maximums or anything about one take, so they can’t be traced back to anyone. We never share a take, or anything about one contributor, with Claude. We keep downloaded copies only as long as we need them, and delete them no later than 3 years after the take was recorded. Takes you haven’t sent stay only on your iPhone.

Apple and Framer handle this data for us as our service providers, which Korea’s PIPA calls entrusted processors: Apple Inc. stores takes and consent records in iCloud, and Framer B.V. hosts this website and stores what you send with the request form. Both handle it on our behalf under their data-processing terms.

9. Where Your Data Travels

9. Where Your Data Travels

Apple processes takes and consent records as our agent under its developer terms. It sends data from the EEA or Switzerland only to countries with adequate protection, or under the EU standard contractual clauses. Framer B.V. (Netherlands) hosts the request form under its Data Processing Addendum. Your data may be stored and accessed outside Canada and Korea, including in the United States. Where your data goes:

Apple Inc., United States (apple.com/legal/privacy/contact). What’s sent: the takes and consent records described in What We Collect. When and how: each time you send a take or agree, over an encrypted connection from your iPhone to iCloud. Why: to store them for SpeechPath Lab. How long: as described in How Long We Keep Things.

Framer B.V., Netherlands, and its hosting providers, which may be outside the EU (legal@framer.com). What’s sent: your email, your Contributor ID or take IDs, and any details you add. When and how: when you send the request form, over an encrypted connection. Why: so we can act on your request and reply. How long: until 1 year after your request is closed.

You can refuse either transfer. Takes you don’t send stay only on your iPhone, though SpeechPath Lab can’t use them, and you can email privacy@myungjoon.com instead of using the form.

10. How Long We Keep Things

10. How Long We Keep Things

Takes stay in SpeechPath Lab’s database until you delete them or withdraw your consent. We destroy each take, and every copy of it, when it’s no longer needed for building lip guides, and in any case no later than 3 years after it was recorded. If a take’s purpose ends sooner, for example because its phrase or language is no longer used, we destroy it within a year of that. Rejected takes follow the same limits, since a take that wasn’t used at first may be used later. We follow a written procedure for this: every quarter, we check SpeechPath Lab’s database and our computer for takes past these limits and destroy them, and we log only how many were deleted. Consent records aren’t takes: they’re kept to show that consent was given and, where it was, withdrawn. Consent records are kept for 5 years after your consent ends or your last take is deleted, whichever is later, then deleted. Requests sent with the request form, and our replies, are deleted 1 year after the request is closed.

12. Your Rights

12. Your Rights

Depending on where you live, including under the GDPR, UK GDPR, Korea’s PIPA, Canada’s PIPEDA and British Columbia’s PIPA, and possibly California’s CCPA/CPRA, you may have the right to access, correct, delete or move your data, to restrict or object to its use, to withdraw consent, and to complain to your local data protection authority, such as Korea’s Personal Information Protection Commission (PIPC) or the Office of the Information and Privacy Commissioner for British Columbia (OIPC). Most of these you can do yourself in SpeechPath Lab. To get a copy of your data, use Export My Data in Settings. It makes a file with every take on your iPhone, every take you’ve sent that’s still in SpeechPath Lab’s database, your consent records, your consent PDF while your consent stands, your signature’s pen strokes, your language, accent and age check, and your Contributor ID. If you’ve deleted SpeechPath Lab, reinstall it on an iPhone signed in to the same Apple Account to export what’s still in the database. We don’t send copies by email, because only the Apple Account on your iPhone can read your data, so the app is how we know it’s really you. For anything else, use the request form or email privacy@myungjoon.com. Requests are matched by Contributor ID or take ID, as described above. If you use the form, we use what you send only to act on your request and reply. We usually reply within a few days, and always within 2 weeks.

13. Children

13. Children

SpeechPath Lab is only for people 18 or older. Before you can record, Apple shares the age range on your Apple Account with SpeechPath Lab, and recording is off unless that range is 18 or over. If we learn that someone under 18 has sent takes, we delete them and every copy without waiting for a request. A parent or the contributor can also delete them on that iPhone with Delete My Recordings in Contributions, or send the Contributor ID or take IDs with the request form or to privacy@myungjoon.com.

14. Keeping Things Secure

14. Keeping Things Secure

Takes travel and are stored with Apple’s iCloud encryption, and CloudKit’s access rules keep each person’s takes private to them and the developer. No system is perfectly secure, and we’ll tell you and the relevant authority about a breach that affects you where the law requires it.

15. Changes to This Policy

15. Changes to This Policy

If a change affects what we collect, who sees it, what it’s used for, how long it’s kept or where it goes, we raise the consent version and SpeechPath Lab asks you to agree again before you record. Other changes are listed with their date at the top of this page. The current consent version is 1. Once there are earlier versions, they’ll be kept online with the date each one applied.

16. Get in Touch

16. Get in Touch

Questions or requests: privacy@myungjoon.com.

Copyright © 2025-2026 Myung Joon Kang.

All rights reserved.

Made with ❤️ in Vancouver and Seoul.

Copyright © 2025-2026 Myung Joon Kang.

All rights reserved.

Made with ❤️ in Vancouver and Seoul.

Copyright © 2025-2026 Myung Joon Kang.

All rights reserved.

Made with ❤️ in Vancouver and Seoul.