Privacy Policy

The short version

• You can browse and listen without an account. Nothing is saved until you sign in.
• Signing in stores your email address, your saved words, your review schedule, the shows you follow, and where you stopped in each episode.
• Nothing is sold, and nothing is shared with advertisers.
• Machine-generated transcripts are shared between everyone using Echo. They contain the episode's speech, never anything about you.
• You can have your account and everything in it deleted by emailing me.

Who runs Echo

Echo is built and run by one person. There is no company behind it and no staff. When this page says “I”, that is who it means, and support@echopodenglish.com reaches me directly.

What is stored when you sign in

You sign in either with Google or by asking for a one-time link sent to your email address. Either way an account is created, and that account holds:

• Your email address, and whether it has been confirmed. This is the only piece of identity Echo needs.
• Your name and profile picture, if you signed in with Google and Google supplied them. Echo asks Google for nothing beyond your sign-in identity, email address and basic profile.
• Sign-in records: a stored session so you stay signed in, and for Google sign-in the tokens Google issues. The session identifier is stored as a one-way hash, so the value in the database cannot be used to sign in as you.
• Email sign-in links, stored only as a hash, valid for fifteen minutes and usable once.

What is stored as you use it

This is the study data — the part the app exists to keep:

• Saved words. The word, its pronunciation, its English meaning, its translation, the sentence you found it in, which episode that was, the timestamp, and the start and end of the short audio clip around it.
• Review scheduling. How many times you have reviewed a word, how easy you found it, when it is next due, and how often you have forgotten it.
• Followed shows. The feed address, title, host and cover art of each podcast you add, and when you added it.
• Playback position. Where you stopped in each episode, so another device can pick it up.
• Everything else the app keeps locally, stored as-is: your listening streak and total listening time, your vocabulary estimate, your language setting, and cached transcripts and translations for episodes you have opened. These sit in a general-purpose store rather than in tables of their own.
• Feature counts. How many caption minutes and translations you have used this month, and which episodes and languages you have already been counted for, so that opening the same episode twice is never counted twice.

Some of this is also kept in your browser's local storage, which is how the app works before you sign in and how it keeps working while the network is bad. Signing out clears it from that browser.

What is not stored

Echo does not record your IP address or your browser's user agent against your account, does not keep a history of what you listened to beyond the position in each episode, and does not build a profile of you for advertising. Cloudflare, which hosts the app, necessarily handles your IP address in order to deliver requests, as any web host does.

Podcasts, and where the audio comes from

Podcasts are public. Echo reads a show's RSS feed and plays the audio files that feed points at, both published by the podcast's own publisher on their own servers. Echo does not host the shows and has no relationship with their publishers.

Feeds, cover art and audio are fetched by Echo's server rather than by your browser. That is a practical decision — many podcast hosts are slow or unreachable from networks here — and it has a privacy consequence worth stating: the podcast host sees a request from Cloudflare, not from you.

Why a copy of the audio is stored

When you ask Echo to generate captions for an episode that ships none, a copy of that episode's audio is written to Cloudflare R2, Cloudflare's file storage. This is what makes transcription work at all: some podcast hosts refuse repeated or unusual requests, and a transcription that stops halfway because the source stopped answering has still cost money. Working from a stable copy also means a retry is free rather than a second charge.

The copy is of the publisher's episode. It is stored under the episode's own identifier, not under your account, and it contains nothing about you.

Other companies that see something

Echo uses outside services for the things it cannot do itself. In each case, what they receive is listed here.

• Cloudflare hosts the app, the database and the file storage, and serves every request. If Cloudflare Web Analytics is switched on, it counts page views without cookies and without identifying visitors.
• Deepgram generates captions. It receives the episode's audio, or the address to fetch it from. It also receives your own voice recording when you use speaking practice — a few seconds of you saying a sentence, sent so it can be turned into text and compared. That recording is not stored by Echo.
• DeepSeek handles translation, word definitions and the AI conversation. It receives transcript lines to translate; a word plus the sentence it appeared in, for a contextual definition; and, in a conversation, your messages together with the episode title, show name, your position in it and the transcript lines around that moment. Your email address and account are not sent.
• Resend delivers email. It receives your email address and the contents of the message — a sign-in link or a welcome note.
• Google handles Google sign-in, and only that. It tells Echo your Google account identifier, email address and basic profile.
• Apple supplies podcast search and charts through the iTunes Search API. It receives the search terms you type and the category you browse, forwarded by Echo's server rather than by your browser.
• Free Dictionary API returns English definitions. It receives the single word being looked up.
• Youdao supplies recorded pronunciations. This one your browser loads directly, so Youdao sees the word and your network address. Youdao is based in mainland China.
• WeChat takes the payment for Echo Pro, if you buy it. That happens inside WeChat, between the two of us — the same as paying any other person. Echo has no checkout, receives no payment details and stores none. What reaches Echo is me switching Pro on for an account by hand.

None of these companies receive your saved words, your progress or your account contents, and none of them are paid in data. What they get is what is needed to answer the request in front of them.

Echo's code also contains connections to services that are switched off and have never been used — a speech provider and a payment processor, both from earlier plans. Neither is listed above, because a company that has received nothing is not somebody your data went to. They are named here only so that finding them in the code isn't a surprise.

Cookies

Echo sets four cookies, all of them functional. There are no advertising cookies.

• echo_session keeps you signed in. It lasts thirty days, and is readable only by the server.
• __Host-echo-device is a random identifier for the browser, kept for a year. It is how the rate limits tell one browser from another, which is what stops a script running up the transcription bill. It is not tied to your account and not used to track you between sites.
• echo_oauth_state, echo_oauth_verifier and echo_oauth_next exist only during a Google sign-in — a few seconds — and are what stop somebody else's sign-in from being substituted for yours.

Where the data lives

Accounts and study data are held in Cloudflare D1, Cloudflare's database. Audio copies and generated transcripts are held in Cloudflare R2, its file storage. Both run on Cloudflare's network outside mainland China, and the app is served from there.

Echo is used mostly from mainland China, so this is worth being direct about: your data is stored abroad, and the services listed above are reached from outside the country. The one exception is Youdao, a Chinese provider, used only to play the pronunciation of a single word.

How long things are kept

• Your account and study data are kept until you ask for them to be deleted.
• Sign-in sessions expire after thirty days. Email sign-in links expire after fifteen minutes.
• Feature counts are kept per week and per episode, and are what make a retry free rather than a second charge.
• Shared transcripts, shared translations and stored audio copies are kept indefinitely, because they belong to an episode rather than to a person and deleting them would make the next reader pay to regenerate them. There is no automatic clean-up of this storage yet.

Deleting your account

There is no delete button in the app yet. Email support@echopodenglish.com from the address you sign in with and I will delete the account by hand, along with your saved words, review schedule, followed shows, playback positions, settings, usage counts and any purchase record. That is everything held against your account, and it is done within thirty days of you asking — usually the same week, but a promise with a date on it is worth more than one without.

Signing out clears Echo's data from that browser, which is worth doing on a shared computer.

Your data is not sold

Echo does not sell personal data, does not share it with advertisers or data brokers, and does not use it to train anyone's models. The services above act on Echo's behalf to answer a specific request.

Children

Echo is not designed for children and is not directed at anyone under 13. If you believe a child has created an account, email support@echopodenglish.com and I will delete it.

Changes to this page

When this page changes, the date at the top changes with it. If a change materially affects what is collected or who receives it, I will say so by email to signed-in accounts before it takes effect. Small corrections will just appear here.

Questions

Ask about anything on this page, or about what is held against your account, at support@echopodenglish.com. It reaches me directly.