Legal
Privacy Policy
Speekl is a speaking app, so this policy has to be specific about your voice. It says what we collect, which companies process it on our behalf, how long we keep it, and what you can ask us to do with it.
Last updated
1.Who we are
Speekl is operated by Speekl, LLC, a limited liability company registered in Delaware, United States, at 1111B South Governors Ave, STE 7418, Dover, DE 19904, USA. We are the controller of the personal data described here.
For anything in this policy — including a request to see or delete your data — write to hello@speekl.com.
This policy covers the Speekl AI tutor: the website, the practice sessions, and the account behind them.
2.What we collect
Before you have an account
The interview that introduces Speekl runs before an account exists. It is stored against an anonymous session token kept in your browser, and it holds the answers you give: the language you want to learn, your native language, why you are learning, what you have tried before, how you rate your own level, how much time you want to commit, and what you want in a tutor. If you take the spoken level check, we store the transcript of what you said and the level we estimated from it.
If we offer to email you a link back to an unfinished interview and you accept, we store that email address for that purpose. It does not create an account.
We also record the `utm_*`, `ref`, `gclid` and `fbclid` parameters that were on the link you arrived through, and your browser's user-agent string, so we can tell which channels bring people to Speekl.
Your account
Your email address, which is also how you sign in — we email you a short code rather than asking for a password. Your name, time zone and profile photo if you provide them.
What you produce by practising
- Text transcripts of your practice sessions, turn by turn — what you said and what the tutor replied.
- The corrections and pronunciation notes the tutor generated from those turns.
- Pronunciation scores for each exercise attempt and for the lesson sentences we score, your progress on each sound, the pronunciation checks you take, and texts you save to practise.
- Your accent preference (American or British English) for each language that has one.
- If you turn on recording history, your pronunciation recordings — see Recording history.
- Short notes your tutor keeps about things you mention, such as an upcoming trip or your job — see What your tutor remembers.
- Vocabulary you save, flashcard decks and review history.
- Your learning plan, goals, progress, streaks and session history.
- The custom tutors you create, including the description you wrote and the avatar generated from it.
Billing
Your plan, its status, and the identifiers that connect your account to our payment processor. Payment happens on Stripe's own pages — card details never reach Speekl's servers and we never store them.
Technical
Each time you sign in we store the IP address and user-agent of that session alongside its token, so a session can be recognised and revoked. Our servers keep operational logs.
Cookies and browser storage are covered separately in the Cookie Policy.
3.Your voice and audio
This is the part of the policy most worth reading, so it is written plainly.
By default, we don't keep recordings of your voice. Audio from your microphone is processed to transcribe it, to let the tutor respond and to score your pronunciation, and then it is discarded. We keep the text and the scores. If you turn on *recording history*, we keep your pronunciation recordings for 90 days so you can replay them — you can turn it off and delete them at any time.
Where the audio goes
There are three paths. Which of the first two a session uses depends on the mode you are in; the third is pronunciation scoring:
- Through our servers. Your audio reaches Speekl, is passed straight to OpenAI for transcription, and is discarded once the text comes back. It exists in our server's memory for the length of that request and is never stored.
- Straight to the voice provider. For real-time conversation your browser connects directly to ElevenLabs using a short-lived token we issue. That audio never touches Speekl's servers at all — it goes from you to the provider.
- Pronunciation scoring. When you practise pronunciation — and, for English learners on Voyager or Immersion, for a sample of what you say in lessons — short clips of your speech go to Speekl's servers and are forwarded to Microsoft (Azure AI Speech) for American English or Speechace for British English, together with the text you were saying. Microsoft processes them in the United States and Speechace in the European Union, and both keep nothing: logging is disabled on our Microsoft service, and Speechace is contracted to zero-day retention. Neither uses them for training.
Lesson scoring is on by default for English learners on Voyager or Immersion, so the tutor can help with the sounds that matter for you. You are told in the lesson lobby, and you can turn it off at any time with Score my pronunciation during lessons in pronunciation settings; capture stops immediately.
Your voice is used to score your pronunciation — never to identify you. We do not create voiceprints or any biometric identifier.
Audio you hear from Speekl — a tutor speaking, a word pronounced — is synthesised speech, not a recording of a person. Pronunciation clips we generate are cached in our file storage so the same word does not have to be regenerated. They contain no recording of you.
Speekl needs microphone access to work at all. Your browser asks before granting it, and you can withdraw it at any time in your browser's site settings — the text side of the product keeps working without it.
4.Recording history (optional)
Recording history is off unless you turn it on, on its own screen, in pronunciation settings. It lets you replay your own attempts and hear how you have changed.
- What is kept: the recordings of your scored pronunciation exercises (words, drills, texts, pronunciation checks, accent coach lessons) and of the lesson sentences we score. Recordings we could not score are not kept.
- How long: 90 days from each recording, then it is deleted automatically.
- Where: in our Amazon Web Services storage, encrypted.
- Who can access it: only you, for playback. It is not used to train models, not shared with anyone, and not reviewed by our staff — support can listen only if you give explicit permission for a specific request.
- Turning it off: stops new recordings from being kept immediately, and offers to delete everything already kept at once. Otherwise each recording is still deleted 90 days after it was made. Deleting your account deletes all of them.
We record when you turned it on and which version of this policy you saw.
5.What your tutor remembers
To make lessons feel like talking to someone who knows you, your tutor keeps short notes about things you mention in lessons and in the chat. For example, it might note that you are going to Spain next week, so it can ask how the trip went.
- What is kept: a short title and one sentence of detail, a date for things that happen on a date, and which tutor and lesson it came from. The tutor is instructed not to note health, religion, politics, sexuality, money or private details about other people unless you ask it to.
- How it is used: at the start of a lesson the notes are added to the tutor's instructions, which are sent to the AI providers that run the lesson (see Who processes it for us). They are not used to train models and not shown to anyone else.
- Seeing and deleting it: all of it is listed on What your tutor remembers, where you can delete any note or all of them. Deleting there is permanent.
- How long: while your account exists. When the tutor itself decides a note is out of date, it stops using it, but the note is kept until you delete it there or delete your account. Deleting your account deletes all of them.
6.How we use it
- To run practice sessions — transcribe what you say, generate the tutor's reply, and speak it back.
- To correct you usefully, which means the tutor needs your level, your goals and what you have got wrong before.
- To build and adjust your learning plan, track progress and keep your vocabulary.
- To score your pronunciation, show you which sounds to work on, and — if you turned on recording history — let you replay your recordings.
- To meter practice minutes against your plan's allowance.
- To take payment, and to email you about your account, your sign-in codes and your subscription.
- To keep the service secure, debug failures and prevent abuse.
- To understand which channels bring people to Speekl, using the attribution parameters described above.
We do not sell your personal data, and we do not use your practice conversations to train our own models or anyone else's. Our providers process what we send them to return a result to us, under their agreements with us.
7.Legal bases
Where the GDPR or UK GDPR applies to you, we rely on:
| Basis | What it covers |
|---|---|
| Performance of a contract | Running your account, your practice sessions and your subscription — the service you signed up for, including scoring the pronunciation exercises you choose to do. |
| Legitimate interests | Keeping the service secure and working, understanding how people find and use it, and preventing abuse. Scoring a sample of your lesson speech for pronunciation feedback, which is on by default for English learners on paid plans and which you can turn off at any time. |
| Consent | Microphone access, the optional recovery email offered mid-interview, and recording history. You can withdraw any of them. |
| Legal obligation | Keeping the records of payments that tax and accounting rules require us to keep. |
Performance of a contract
- What it covers
- Running your account, your practice sessions and your subscription — the service you signed up for, including scoring the pronunciation exercises you choose to do.
Legitimate interests
- What it covers
- Keeping the service secure and working, understanding how people find and use it, and preventing abuse. Scoring a sample of your lesson speech for pronunciation feedback, which is on by default for English learners on paid plans and which you can turn off at any time.
Consent
- What it covers
- Microphone access, the optional recovery email offered mid-interview, and recording history. You can withdraw any of them.
Legal obligation
- What it covers
- Keeping the records of payments that tax and accounting rules require us to keep.
8.Who processes it for us
Speekl is built on other companies' infrastructure. These are all of them, and what each one receives:
| Provider | What it does for us | What it receives |
|---|---|---|
| OpenAI | Transcribes speech, generates the tutor's conversation and its corrections | Practice audio, session transcripts, and the parts of your learning profile the tutor needs |
| ElevenLabs | Speaks the tutor's replies, and carries the real-time voice conversation | Your microphone audio and the conversation text, in the direct path described above |
| Anthropic | Writes the profile for a custom tutor you create | The description you wrote for that tutor |
| Generates the avatar image for a custom tutor | The description of that tutor | |
| Stripe | Takes subscription payments and hosts the checkout and billing pages | Your email address and the payment details you enter on Stripe's pages |
| Resend | Sends account email — sign-in codes and subscription notices | Your email address and the contents of that email |
| Microsoft (Azure AI Speech) | Scores the pronunciation of short recordings of American English | The recording and the text being read |
| Speechace | Scores the pronunciation of short recordings of British English | The recording, the text being read, and an anonymised user identifier |
| Amazon Web Services | Stores generated files (pronunciation audio and tutor avatars) and, if you turn on recording history, your pronunciation recordings for 90 days | The generated files, and your recordings if recording history is on |
| Expo | Delivers push notifications, if you use the Speekl mobile app | Your device's push token and the notification text |
OpenAI
- What it does for us
- Transcribes speech, generates the tutor's conversation and its corrections
- What it receives
- Practice audio, session transcripts, and the parts of your learning profile the tutor needs
ElevenLabs
- What it does for us
- Speaks the tutor's replies, and carries the real-time voice conversation
- What it receives
- Your microphone audio and the conversation text, in the direct path described above
Anthropic
- What it does for us
- Writes the profile for a custom tutor you create
- What it receives
- The description you wrote for that tutor
Google
- What it does for us
- Generates the avatar image for a custom tutor
- What it receives
- The description of that tutor
Stripe
- What it does for us
- Takes subscription payments and hosts the checkout and billing pages
- What it receives
- Your email address and the payment details you enter on Stripe's pages
Resend
- What it does for us
- Sends account email — sign-in codes and subscription notices
- What it receives
- Your email address and the contents of that email
Microsoft (Azure AI Speech)
- What it does for us
- Scores the pronunciation of short recordings of American English
- What it receives
- The recording and the text being read
Speechace
- What it does for us
- Scores the pronunciation of short recordings of British English
- What it receives
- The recording, the text being read, and an anonymised user identifier
Amazon Web Services
- What it does for us
- Stores generated files (pronunciation audio and tutor avatars) and, if you turn on recording history, your pronunciation recordings for 90 days
- What it receives
- The generated files, and your recordings if recording history is on
Expo
- What it does for us
- Delivers push notifications, if you use the Speekl mobile app
- What it receives
- Your device's push token and the notification text
We may also disclose data where the law requires it, or to a buyer if the business is sold — in which case this policy travels with it until you are told otherwise.
9.Where your data goes
Speekl, LLC is a United States company, and the providers above are United States companies. If you are in the European Economic Area, the United Kingdom or Switzerland, using Speekl means your data is transferred to and processed in the United States, under the Standard Contractual Clauses or an equivalent safeguard in our agreements with those providers.
Pronunciation scoring for British English is the exception: Speechace processes those recordings in the European Union.
10.How long we keep it
We keep your account and your learning data for as long as your account exists. That is deliberate: your transcripts, corrections and progress are what let the tutor pick up where you left off, and deleting them on a timer would quietly make the product worse.
- Practice audio — not stored, so nothing to retain. See Your voice and audio.
- Pronunciation recordings — only with recording history on, and for 90 days from each recording. See Recording history.
- Pronunciation scores and sound progress — kept while the account is open, like the rest of your learning data.
- Transcripts, corrections, vocabulary and progress — kept while the account is open.
- What your tutor remembers — kept while the account is open, unless you delete it sooner. See What your tutor remembers.
- Interviews that never became an account — kept so you can come back and finish; ask us and we will delete one.
- Sign-in sessions — expire after 30 days, and are revoked immediately when you log out.
- Payment records — kept as long as tax and accounting rules require, which is longer than the account.
When you ask us to delete your account we act on it — see the next section for how, and for what to expect.
11.Your rights
Depending on where you live, you can ask us to give you a copy of your data, correct it, delete it, restrict or object to how we use it, or hand it over in a portable form. You can also complain to your local data protection authority.
To exercise any of these, email hello@speekl.com from the address on your account. We will respond within 30 days, and we do not charge for it.
Deletion is handled by us, on request — there is no self-service delete button in the app yet. Emailing us is the way to do it, and it is not a lesser route: a request removes the account and the learning data behind it, including pronunciation scores and any kept recordings. We are building the in-app version, and this section will change when it ships.
Some things survive a deletion request because the law says they must — payment and tax records in particular. We will tell you what was kept and why.
12.Security
Traffic to Speekl is encrypted in transit. Your session lives in a cookie your browser's scripts cannot read, sent only to Speekl's own domains. Sign-in is a one-time emailed code, so there is no password of yours for us to lose. Sessions expire and can be revoked.
No service is perfectly secure, and we would rather say so than imply otherwise. If you believe your account has been accessed by someone else, email us and we will revoke its sessions.
13.Children
Speekl is not intended for children under 13, and we do not knowingly collect their data. If you believe a child has given us personal data, email us and we will delete it.
14.Changes to this policy
When this policy changes, the date at the top changes with it. If a change materially affects what we do with your data, we will tell you by email or in the app before it takes effect.
15.Contact
Speekl, LLC, 1111B South Governors Ave, STE 7418, Dover, DE 19904, USA. Email hello@speekl.com.