_Last updated: 30 September 2026_
This notice explains how LopoRail handles voice input when you log sightings by voice through an assistant, and how the on-device sound identification feature uses your microphone.
1. Voice logging through assistants
1.1 You can log sightings by voice through an assistant you already use — for example Siri Shortcuts, Alexa, or an automation service such as IFTTT or Zapier. This is off by default: nothing is accepted by voice until you switch voice logging on in your spotting preferences.
1.2 The assistant captures and understands your spoken words. That audio is handled by the assistant's own provider under its privacy terms — we do not receive or store the audio. We receive only the text the assistant sends us.
1.3 Your assistant connects using a dedicated access token that can only create sightings. It cannot read, export or delete anything. You can revoke the token at any time in your spotting preferences, which immediately ends that access.
1.4 To protect against misheard numbers and station codes, we may read the sighting back for confirmation before saving it, and repeat requests are de-duplicated so a retried turn never logs twice.
1.5 When an assistant logs a sighting, we record that it came from voice and which assistant sent it, alongside the normal sighting details (train, station, time).
1.6 In the mobile app — the app supports hands-free voice logging. It uses your phone's speech recognition to turn what you say into text, and your phone's speech service processes that speech under the phone maker's own terms. We receive only the text through our voice logging endpoint — we do not receive or store your audio. The app declares its microphone permission, and voice logging is off until you enable it; you can turn it off at any time.
2. Sound identification (microphone)
2.1 The "identify by sound" feature asks for microphone permission in your browser and captures a few seconds of audio to analyse a passing train.
2.2 The analysis runs entirely on your device — the sound is matched against our profile set in your browser. No audio is transmitted to us or stored. Deny the permission and the feature simply does not run.
2.3 The result is a suggestion with a confidence level. If you choose to attach it to a sighting, we save the suggested result — not the audio — alongside that sighting.
3. Accuracy
3.1 Speech recognition and sound identification can be wrong, especially with numbers, station codes and similar trains. Treat both as aids, not facts; the confirmation step in §1.4 exists for this reason.
4. Your choices
4.1 Turn voice logging on or off at any time in your spotting preferences; voice logging is off until you enable it.
4.2 Revoke assistant tokens at any time — the Assistant panel lists every token you have issued.
4.3 Microphone use is entirely under your browser's permission control; we never access the microphone without it.
4.4 Ask us about the voice-related data we hold, or to remove it, at our contact page.