Support & FAQ

Echolate works like a consecutive interpreter: it waits until a phrase is finished, then recognizes, translates, and voices it. That pipeline takes 2–5 seconds per phrase. The upside is accuracy — the full phrase gives the recognizer and translator real context, instead of guessing word by word.

The original soundtrack keeps playing the whole time, so you never lose the thread — the translation simply trails slightly behind, like a live interpreter at a conference.

Echolate deliberately has no backend. Your audio goes straight from your browser to Google's APIs, and the only way to do that without routing your data (and your money) through someone else's server is to use your own key. It also means you pay Google's raw prices with no markup.

Google's free monthly quotas cover roughly an hour of casual use. Beyond that, continuous translation costs about $1.5–2 per hour, most of it Speech-to-Text. See the cost table in the setup guide, and set a budget alert so there are no surprises.

Google's Speech-to-Text API allows only a primary language plus a few alternatives per request. In Auto mode Echolate currently asks it to choose among English, Ukrainian, Russian, and German.

The fix: if the audio is in any other language, select the source language explicitly in the popup instead of Auto. Explicit selection supports 85+ locales and is also slightly more accurate and faster than Auto mode.

Echolate estimates the speaker's pitch and requests a matching male or female voice from Google Text-to-Speech. However, not every language has TTS voices of both genders — for some languages Google offers only one. In that case Echolate uses the voice that exists, even if it doesn't match the speaker.

Two different limits, both imposed by the browser:

DRM-protected audio (Netflix and some other paid streaming services) may be blocked from tab capture by the platform's content protection, so there is no audio for Echolate to hear.

Browser system pages — chrome:// pages, the Chrome Web Store, and other internal pages — cannot be captured by any extension. Regular websites, YouTube, streams, and podcasts all work.

Work through this list in order:

1. All three APIs are enabled on the project: Speech-to-Text, Translation, and Text-to-Speech. This is the most common cause — links in the setup guide, step 2.
2. Billing is enabled on the project. Google rejects requests from projects without an active billing account, even inside the free quota.
3. The key's restrictions are right. If you restricted the key, it must allow exactly those three APIs, and it must not have a website/HTTP-referrer restriction.
4. The key was copied whole — no spaces or missing characters.
5. The right project was selected in the Cloud Console when you enabled the APIs — it's easy to enable them on a different project than the key belongs to.

You're only billed while requests are being made, so simply not using Echolate means no new charges. For extra certainty, any of these works (from softest to hardest):

• Remove the API key from the extension's settings (or uninstall the extension).
• Set a budget alert in Billing → Budgets & alerts so you're emailed before spending grows.
• Disable the three APIsin the Cloud Console (APIs & Services → each API → Disable) or delete the API key — requests stop immediately.
• Delete the whole Google Cloud project — nothing on it can ever bill again.