Privacy Policy

This page says what Read to Me does with your text and your data. It describes what the site actually does today, and we update it when that changes. Effective 2026-11-16. Questions: support@readtomeai.com.

Device voices: we never receive the text

Device voices are the voices built into your browser or phone. When you use one, your own device speaks the text. Our code does not send that text to our servers.

Some browsers also offer online voices, for example Chrome's online Google voices, which send the text they speak to the browser maker's own servers. That happens inside your browser and we cannot see or control it. Chrome's developer documentation describes such a voice as a remote network resource. If you do not want that, pick a voice your browser does not mark as online, or check your browser's privacy settings.

AI voices: each sentence goes to DeepInfra

When you choose an AI voice, the page sends the sentence it is about to read (a very long sentence is cut into pieces, at most 600 characters per request) and the voice name to our server. Our server passes the text on to DeepInfra, which runs the Kokoro-82M speech model and sends back an MP3. Your browser plays it. We do not save the audio or the text. DeepInfra receives the text and the voice name only: we do not forward your IP address, cookie or device hash to it.

DeepInfra is a separate company with its own policies. These are the statements on its pages when we last checked them (2026-11-16):

Those are DeepInfra's statements. We cannot inspect its systems, and its pages can change. If you do not want your text processed by DeepInfra, use a device voice: our code does not send that text anywhere.

What our server stores

Each time an AI voice generates audio for a sentence, we add one row (playing the same sentence again with the same voice on the same day reuses that row) to a table called ai_usage in our database (hosted on Cloudflare). It is how we count your free AI minutes and apply the daily limits. A row holds:

dayThe UTC date of the request.To count each day's use; the allowance resets at midnight UTC.
cookie_idThe random ID from the rtm_aid cookie (see Cookies below).To count your own free AI minutes.
fpA device fingerprint hash. Your browser combines its user-agent string, language, screen size and colour depth, time zone and CPU core count, hashes them (SHA-256) and sends the first 32 characters. We store only that hash.So the daily allowance still counts if you clear your cookies. Used only for the free-allowance limits.
ipYour IP address as seen by Cloudflare.For the daily cap on one network (a shared Wi-Fi counts together).
voiceWhich AI voice you picked.Part of the record of what was generated.
text_hashA short one-way hash (16 characters) of the sentence. The sentence itself is not stored.To recognise a sentence you play again with the same voice, so a replay is not counted twice.
charsHow many characters the sentence has.Usage and cost accounting.
est_seconds, secondsEstimated and actual length of the audio.To count your free minutes.
statusreserved while generating, done when finished.A failed generation is removed and is not counted.
cost_microWhat the generation cost, in millionths of a US dollar.To keep the daily spending cap on free AI.
created_atThe time of the request.Used by the cleanup that deletes old rows.

We do not store the text you read or the audio we generate. These records are not sent to Google Analytics or Microsoft Clarity. We use them only for the allowance and abuse limits described here.

How long we keep it. A cleanup job runs once a day and deletes ai_usage rows early enough that none stays longer than 30 days.

A second table, daily_budget, holds one row per day: the date and the total spent on free AI audio that day. It contains no personal information. The same job deletes rows older than 400 days.

To notice when something breaks, each time the page asks how much free time you have left we also note the hour and the first 8 characters of a one-way hash of your cookie ID (table ops_hits). It is used only to count how many visitors we have and is deleted within 2 days.

For each AI request our server also writes one log line with the result code, a short result class and how long it took. The log line does not include the text, your IP address or your cookie ID.

What stays in your browser

These items live in your browser's local storage on your device. Our server does not receive them. You can delete them any time by clearing this site's data in your browser settings.

rtm:reader:v1Your reading progress for up to 5 texts: the text itself, a title taken from its first sentence, the sentence you reached and when you last read it.So the Continue card can take you back to where you stopped. The text is stored here, on your device only.
rtm:reader:prefsThe voice and speed you chose.To remember your settings.
rtm:consentWhether you pressed Accept or Reject on the cookie banner.So the banner does not ask again. Kept until you clear site data.
rtm.sent.purchaseThe order numbers of purchases already reported to Google Analytics (the most recent 50).So one purchase is reported once. Paid plans are not on sale yet, so nothing is written to it today.

PDF files

When you choose a PDF, the page opens it in your browser and pulls the text out there, using the open-source pdf.js library. The file is not uploaded: our code never sends the file to our servers. With an AI voice, the text taken from it goes to our server one sentence at a time, exactly as described under AI voices above. With a device voice our code does not send the text to us; what your browser's own online voices do with it is covered under Device voices above.

To let you pick up where you stopped, the text taken from the PDF is saved in your browser under rtm:reader:v1 (see What stays in your browser), together with a map of which sentence starts each page and the page count. The file itself and its file name are not saved. A long PDF can be too big for your browser's storage; if the browser refuses the save, the page keeps working and the place is simply not remembered.

The file name is not sent to Google Analytics: the events pdf_upload (number of pages) and pdf_rejected (a reason code) carry no name or text. In Microsoft Clarity, the card that shows the file name and the text shown while a PDF is read are masked, so they appear as blocks in recordings and are not captured.

Cookies

First party means we set the cookie on our own domain. The Google and Microsoft cookies are written by their scripts; where a lifetime is not given here, it is set by Google or Microsoft and described in the documentation we link to.

rtm_aidUs (our server)A random ID that counts your free AI minutes. It is set the first time a page with the reader checks your allowance, even if you never play an AI voice.365 daysFirst party, HttpOnly
at_ccUs (our front door)The two-letter country code Cloudflare works out from your IP address. The page reads it to decide whether to show the cookie banner.7 daysFirst party
rtm_ftUs (our front door)A first-visit record: the page you landed on, the website that sent you (its domain only), any utm_source, utm_medium or utm_campaign in the link, your country and the time. It is set on your first page view, except in a European Economic Area country (the EU, Iceland, Liechtenstein and Norway), the United Kingdom or Switzerland or when we cannot tell where you are, where it is not set at all. It is sent back to our site with later requests and is not sent to Google or Microsoft.365 daysFirst party, HttpOnly
_ga, _ga_9SGDP38RLKGoogle AnalyticsTell visits and sessions apart.Set by Google, see Google's cookie documentationFirst party cookies written by Google's script
_clck, _clskMicrosoft ClarityLink the pages of one visit into one session replay.Set by Microsoft, see Clarity's cookie listFirst party cookies written by Microsoft's script
CLID, ANONCHK, MR, MUID, SMMicrosoft ClarityListed by Microsoft as Clarity cookies that identify browsers across Microsoft sites. It lists MUID as used for advertising, site analytics and other operational purposes.Set by Microsoft, see Clarity's cookie listThird party (Microsoft)

The Google and Microsoft cookies are only set after you press Accept if you are in a European Economic Area country (the EU, Iceland, Liechtenstein and Norway), the United Kingdom or Switzerland, or if we cannot tell where you are. Everywhere else they are set by default (see Analytics below).

Analytics: Google Analytics and Microsoft Clarity

Google Analytics 4 (measurement ID G-9SGDP38RLK) counts visits and which features are used. These are the events our code sends. None of them contains the text you read, a title or a file name:

We set Google's advertising signals (ad_storage, ad_user_data and ad_personalization) to denied for every visitor. Because Google's script loads from Google, Google receives what any browser sends when it loads a script, such as your IP address and browser details; see Google's privacy policy.

Microsoft Clarity records how pages are used (clicks, scrolling, mouse movement and page views) as replays and heatmaps. We mask the paste box, the text shown while it is being read, and the title on the Continue card, so they appear as blocks in recordings and are not captured. Microsoft receives what Clarity records; see the Microsoft privacy statement and Clarity's cookie list.

Your choice. If you are in a European Economic Area country (the EU, Iceland, Liechtenstein and Norway), the United Kingdom or Switzerland, or if we cannot tell where you are, a banner at the top of the page asks you to Accept or Reject. Until you accept, Google Analytics is told that analytics storage is denied and Clarity is not loaded. Under Google's consent mode, Google's script may still send cookieless pings, which carry no cookie ID but include details your browser sends with any request, such as the page, user agent, screen resolution and IP address (how Google describes it). Your choice is stored in your browser (rtm:consent).

To change your mind, clear this site's data in your browser settings and reload the page: the banner returns. Everywhere else, analytics is on by default and there is no banner or on-page switch yet; your browser's tracking protection or a content blocker will stop the Google and Microsoft scripts.

Children

Read to Me is not meant for children under 13. We do not knowingly collect personal information from them. If you believe a child under 13 has been using the site, write to support@readtomeai.com.

Changes and contact

When we change how the site handles data, we update this page and its effective date. This version is effective 2026-11-16. For any question about this page or your data, email support@readtomeai.com. See also our Terms of service and Refund policy.