TrailEar: Offline Birding Companion for Hacktoberfest Touch Grass Challenge

Описание и применение

TrailEar – автономный помощник для наблюдения за птицами, использующий локальные модели для распознавания звуков и генерации резюме, без подключения к сети.

Задача

Bird‑ID apps require signal, keep users on a screen, and send audio and location to a server, which is impractical for trail walks.

Результат

TrailEar provides offline bird detection and summary, preserving privacy, eliminating network costs, and allowing modular replacement of detection or LLM components.

Рабочий процесс

• 1. User starts TrailEar locally via uvicorn command.
• 2. App captures 3‑second mic windows at 48 kHz mono.
• 3. BirdNET‑Analyzer processes audio, applies confidence threshold and 60‑second cooldown per species.
• 4. When a bird is detected, a quiet cue is emitted via server‑sent events.
• 5. Detections are stored in SQLite.
• 6. After the walk, a local LLM (Ollama or llama.cpp) summarizes the detection list using a static species file, constrained not to add facts.
• 7. User views summary and history through the static HTML/JS frontend.

Доказательства

TrailEar listens through the mic, gives a quiet cue when it detects a bird, and saves the details for a summary after the walk.

Источник

BirdNET-Analyzer on 3-second mic windows (sounddevice, 48 kHz mono), with a confidence threshold and a 60-second cooldown per species so cues stay rare.

Источник

Privacy by design: the server binds to localhost, there is no telemetry, and there are no runtime network calls.

Источник