Echo adds a Listen button to supported reading surfaces — X posts, Grok answers, ChatGPT answers, and Substack articles — and reads them aloud with an on-device neural voice. It can also read selected text through the right-click menu. Speech synthesis and audio playback happen locally in your browser.
What Echo Does Not Collect
Echo never sends, stores, or transmits to any Echo-operated server:
- Post text, article text, AI answer text, selected text, or reading content
- Audio generated from that content
- Twitter/X handles, display names, Substack author names, or profile data
- URLs or page paths you visit
- Personally identifiable information
Network Requests
Echo makes these kinds of third-party network request:
- Voice model download — Hugging Face.The first time you click Listen, after you agree in the consent prompt, Echo downloads the Kokoro voice model (~80 MB). It is cached locally and is not downloaded again unless you clear the cache.
- X post text.To read the full text of an X post, Echo may request it from X/Twitter's public syndication or publish endpoints using the post ID. This request goes to X/Twitter, not to Echo.
- Supported page content — local only. On Substack article pages and ChatGPT conversation pages, Echo reads visible supported content from the page DOM locally so it can synthesize speech. Echo does not send article text or AI answer text to Echo-operated servers or to a third-party text-processing service.
Optional Anonymous Usage Statistics
Echo can collect anonymous usage statistics to help improve the product. This is off by default and only enabled if you explicitly opt in. When enabled, Echo sends only event names, coarse categories, timings and counts, extension version, browser locale, and a rotating anonymous session identifier through Aptabase.
Telemetry never includes post text, article text, AI answer text, selected text, transcripts, handles, names, or URLs. You can turn it off any time in the side panel under Settings.
Permissions
- storage saves local settings such as speed, pronunciation rules, and consent.
- offscreen runs the voice model and audio in a hidden document, which is needed for Chrome MV3 extensions.
- sidePanel shows the player UI.
- contextMenusadds the right-click "Read selection with Echo" entry.
- Host access to x.com and twitter.com injects the Listen button and reads X post content.
- Host access to *.substack.com injects the Listen button and reads article content locally on Substack article pages.
- Host access to chatgpt.com and chat.openai.com injects the Listen button and reads ChatGPT answer text locally from the page DOM.
- Host access to X/Twitter syndication/publish hosts fetches full X post text.
- Host access to Hugging Face hosts downloads the voice model.
- Host access to Aptabase sends anonymous usage statistics only when enabled.
This Website
This website (separate from the extension) uses Umami for privacy-friendly analytics. It sets no cookies, does not fingerprint you, and does not track you across sites — which is why you see no cookie banner here. It records aggregate page views, referrers, and whether the "Add to Chrome" button was clicked. It never identifies individual visitors.
Retention And Contact
Echo stores settings locally in your browser. Uninstalling the extension removes them. Anonymous statistics, if enabled, are retained by Aptabase under their policy.
Questions: hello@echoreadaloud.com