Local Transcribe
中文
Download

Features

Import, transcribe, read, ask and export in one window.

Transcription

Hand an audio or video file to a local model. The work happens on your computer.

  • Import by drag and drop or file picker. The transcript builds up as it runs.
  • Jobs queue up and you can cancel them at any time.
  • Source media is read-only and never reaches a cloud service.
The import area on Home: the heading Home, one line of help text and an Import a file tile

Models

Models download from a curated, measured list inside the app. Arbitrary imports are no longer accepted.

  • The library includes transcription, punctuation and Qwen3 language models.
  • You can set a default model and delete downloaded ones. Deletions go to the system trash.
  • Models load on demand and are reused. The green dot only lights up when a model is actually resident in memory.
  • Idle models unload automatically, after 120 seconds by default. Anything from 20 seconds to 2 hours is available.
The model library showing Qwen3 language models.

Recording

Record inside the app, then transcribe the result locally.

  • Record the microphone and a chosen application together.
  • Each track is transcribed on its own and becomes its own record.
  • No virtual audio driver is installed.
The Recording page: a track list where each track can be an input device or an application, with delay and mute controls, then the total duration and a Start recording button

History and export

Records are grouped by date and open in a reader.

  • Home merges transcription and history. Opening a record shows its text in the reader.
  • The toolbar search matches file names and transcript text, and the reader can search again.
  • Export as TXT or JSON.
The reader: back, find, TXT and JSON export controls at the top, and the text of a completed transcript below

Ask AI

Use a local language model in the reader to explore your transcript.

  • Summarize, extract key points, list action items, or explain a selection.
  • Ask about the transcript and follow up; longer transcripts are processed in sections.
  • Download a language model separately and check generated answers against the original.
The Ask AI panel in the reader, with quick actions and a locally generated summary

Get started easily

Choose your interface language and see model recommendations for your computer on first launch.

  • Recommendations consider your language needs and device resources.
  • The language model is optional; continue without downloading and change settings later.