Self-hosted · Private · Open

Record. Transcribe.
Understand.

AI-powered voice recorder that turns your speech into text and smart summaries. Runs entirely on your own hardware — no cloud APIs, no data leaks.

$ curl rec.dasm.asia/api/v1/transcribe -F file=@meeting.m4a

Uploading... 3.2MB

Processing with Whisper STT...

Generating summary with GLM 4.7...

✓ Done in 12s

{ "status": "completed", "transcription": "..." }

Features

Everything runs
on your hardware

Whisper AI fine-tuned for Russian with 13.78% WER. Runs on your GPU — no OpenAI API needed.

GLM 4.7 Flash generates concise summaries. Custom assistant prompts for different use cases.

Your data never leaves your network. PocketBase + RustFS on your own server. Zero cloud dependencies.

Rust backend. 26MB Docker image. Sub-second API responses. GPU-accelerated inference on RTX 3090.

Flutter app for iOS and Android. Record, upload, read — all from your phone.

No analytics. No tracking. No third-party APIs. Your recordings are yours alone.

How it works

Open the app, tap record. Supports MP3, WAV, M4A, OGG, FLAC — up to 25MB per file.

Audio is transcribed by Whisper and summarized by GLM — all on your GPU, nothing leaves your network.

Get a push notification when done. Read the full transcription and AI summary in the app.