Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

Memoria: Keeping Family Voices Alive with Local Whisper + Gemma

Описание и применение

Memoria – локальный архиватор устной истории, который использует Whisper и Gemma для транскрипции и структурированного резюме голосовых заметок семьи, не отправляя данные в облако.

Задача

Family voice memos contain sensitive personal data that should not be sent to cloud services for transcription and summarization.

Результат

Memoria achieves fully local, privacy-preserving transcription and structured summarization of family voice memos, producing JSON and Markdown archives within ~30 s on a local CPU.

Рабочий процесс

  • Upload or record a family voice memo
  • Run Whisper small (CPU int8) to generate transcript with timestamps
  • Feed transcript to Gemma 3 4B (llama.cpp) to produce structured fields (title, summary, people, places, time references, themes, quotes, follow-up questions)
  • Apply deterministic quote verification to ensure quotes exist in transcript
  • Export results as JSON and Markdown
  • Optionally view via the FastAPI web interface

Доказательства

Memoria runs both speech recognition and language understanding locally. The recording and transcript never leave the machine for inference.

Источник

The bundled 58‑second story is synthetic and fictional, narrated with TTS specifically so that no private family recording is exposed.

Источник

I ran the project rather than stopping at a screenshot or a plausible demo response: Unit tests 6/6 passed, demo audio duration 57.79 seconds, full CLI run 33.36 seconds.

Источник

  • Дата записи: 2026-10-27T09:11:52.202Z