MTranscriber brings the power of OpenAI Whisper to a modern, elegant desktop interface. Zero cloud dependencies, zero data collection, and blazing fast local inference.
Local neural speech transcription with a zero-compromise desktop experience.
Audio is processed exclusively on your CPU or GPU. No network requests are made during transcription. Temporary audio conversion buffers are sanitized immediately after inference.
Instantly search spoken languages by name, native script (e.g. Română), or ISO code. Perform post-transcription translation into Romanian and 30+ languages without re-running audio recognition.
Choose between Tiny (~75MB), Base (~140MB), Small (~460MB), and Medium (~1.5GB) depending on your speed and accuracy requirements. You can also import any custom GGML .bin model file.
Full native bindings to whisper.cpp with Apple Silicon Metal GPU acceleration on macOS and optimized AVX2 SIMD instructions on Windows and Linux.
Stream real-time millisecond-precise timestamped chunks ([00:01.200 → 00:05.100]) alongside a clean, consolidated full transcript with real-time word and segment counters.
Export your transcripts in one click to SubRip Subtitles (.srt), WebVTT (.vtt), Plain Text (.txt), or structured .json. Quick copy button included.
Run MTranscriber on Windows, macOS, or Linux.
Download the installer wizard or standalone portable package:
Static FFmpeg is automatically included or installed via 1-click inside the app.
Run natively with .NET 8 on Apple Silicon or Intel:
Uses macOS native CoreAudio/afconvert engine with zero extra dependencies required.