In short: everything EchoLing AI says and writes for you — translated speech and bilingual subtitles — is generated by artificial intelligence, not by human interpreters. AI translation is very useful, but it can be wrong. For anything important, have a qualified human verify the translation.
1.1 EchoLing AI provides real-time, two-way speech translation. The core capabilities — speech recognition, translation, and speech synthesis — are performed by Google's Gemini real-time translation model (gemini-3.5-live-translate) via the Google Gemini API.
1.2 During a session, your voice (and the other party's voice captured via system audio) is streamed to the Gemini API, which returns translated audio and subtitles to your device. EchoLing AI is an integrator of this model service; we do not train or host the underlying model ourselves.
1.3 The translated voice you and the other party hear is an AI-synthesized voice, not a recording of a human speaker.
2.1 AI-generated translations may contain errors, omissions, mistranslations, ambiguities, or delays, and may not fully capture tone, idiom, cultural nuance, technical terminology, or the speaker's intent.
2.2 Translation quality can vary with audio quality, background noise, accents, speaking speed, overlapping speech, network conditions, and language pair.
2.3 The output is produced automatically and is not reviewed by humans before delivery.
3.1 EchoLing AI is designed for everyday conversations and business communication where participants understand that machine translation is being used and accept its limitations.
3.2 Do not rely on EchoLing AI as the sole basis for decisions in high-stakes contexts, including legal proceedings or contracts, medical consultations and diagnoses, financial or investment commitments, immigration or government procedures, and emergency or safety-critical communication. In such situations, engage a qualified human interpreter or have the content professionally verified.
3.3 Where accuracy matters, confirm key points (numbers, dates, names, commitments) with the other party, and review the bilingual subtitles to cross-check the translation against the original.
4.1 The other party hears an AI-generated voice rather than your natural voice. In many professional and personal contexts it is good practice — and in some contexts may be legally required — to inform the other participants that AI translation is being used.
4.2 If you enable local recording or rely on subtitle archives, you are responsible for complying with applicable recording and consent laws. See the EULA, Section 5.
5.1 Session audio is processed by the Google Gemini API in real time; under Google's paid-tier API terms, your data is not used to train Google's models. We do not store call audio or transcripts on our servers; subtitle history and recordings are saved only on your device. Details are in our Privacy Policy.
6.1 To the extent permitted by applicable law, we are not liable for consequences arising from reliance on AI-generated translations. The applicable warranty disclaimers and liability limitations are set out in the EULA, Sections 10 and 11.
Questions about this disclosure: contact@echolingai.com · echolingai.com
This disclosure may be updated as the product and underlying model services evolve. The effective date above reflects the latest revision.