Hi Product Hunt! I built Voice Rack because my recordings (lectures, meetings, interviews, voice memos) were scattered and unnamed, and I didn't want to upload any of them to a cloud transcription service. Point it at a folder and it transcribes, labels speakers, writes summaries and tags, and lets you ask questions. Every answer cites the recording and the second it came from; click to hear the source. Everything runs locally: the app ships without network permission, and the models live inside the app (about 5 GB). Per-recording features are free; Pro (one-time purchase, 14-day free trial) adds asking across the whole library. Needs Apple silicon and macOS 26. Under the hood, the language model runs on Tempo9, our own production-grade inference engine for Apple silicon and NVIDIA GPUs. Mac developers can use the same engine through the Swift SDK on GitHub (github.com/thinkspread/tempo9). More at thinkspread.com. I'd love feedback, especially on what would make you trust the summaries.

Voice Rack

Offline AI transcription for voice recorders and memos

Offline AI transcription for voice recorders and memos