
The whole pipeline runs on my own machine, so recordings of meetings and interviews never leave my computer โ that alone made it an easy choice over cloud transcription services. Accuracy on Polish and English audio is excellent thanks to the large Whisper model, and GPU acceleration makes it genuinely fast: a one-hour recording is done in a few minutes on a mid-range RTX card. Speaker diarization, the local LLM summaries, and being able to "chat" with a transcript to pull out decisions and action items are the features I use every day. Batch import with automatic transcription and a custom dictionary for names and jargon are big time-savers. Setup was straightforward and it works offline. Review collected by and hosted on G2.com.
It needs a reasonably capable GPU to be quick โ on a CPU-only laptop the long-model transcription is noticeably slower. The local LLM models are a large one-time download. The interface is functional and clear rather than flashy, and a few advanced options take a moment to find the first time. None of this stopped me from adopting it. Review collected by and hosted on G2.com.