TrueScribe transcribes interviews, meetings, lectures, podcasts and videos on your own machine, then gives you the tools to review, correct, translate, search and export what was said.
Local transcription for Windows and Linux.
No account, no upload queue, no waiting for a server.
Drop in an audio or video file, or record a dictation directly. With Pro you can also paste a direct media link or a supported public media page.
TrueScribe runs Whisper locally and picks a model that suits your hardware, downloading it once on first use. CPU and Vulkan runtimes are both supported.
Play back against the waveform, correct the text with undo and redo, name your speakers, translate, and export in the format your work needs.
The same workflow fits very different material.
Keep two speakers apart, correct names and terms once, and quote from a transcript that carries timestamps.
Turn a long recording into text you can skim, search and hand on, without sending the room's conversation anywhere.
Follow the audio next to the text, fix technical vocabulary, and export study notes or subtitles.
Produce show notes, subtitle files and searchable archives from the episodes you already have.
Transcription is the start. The work after it is where most of the time goes.
Timestamped segments sit next to waveform playback with variable speed, so you can check a passage and correct it immediately. Editing supports undo and redo.
Local diarization models detect who spoke when. You can rename speakers, reassign segments and keep the labels consistent across the transcript.
Translate a transcript through an isolated local worker. The translated text stays alongside the original so you can compare or strip it again.
Every transcription is kept in a local SQLite history you can search by text, so an old recording is a query away rather than a folder hunt.
Export as TXT, SRT or WebVTT on the free tier. Pro adds Markdown, DOCX, PDF, JSON and CSV for documents, data work and archives.
Ask questions about a transcript or generate a summary. You choose whether that runs on a local model or on a cloud provider you connect yourself.
TrueScribe can run transcription, speaker detection, translation, AI, editing, history and search on your machine. You decide when a feature may use a cloud model or remote endpoint.
Media files are decoded and transcribed locally. There is no upload step and no processing account.
Whisper and diarization models are fetched on first use and then reused offline.
If you choose a cloud model for transcription, translation, AI chat or another feature, the content needed for that operation leaves your device. The local privacy guarantee does not apply to it.
Transcribing a link, downloading a model, checking your licence and error reporting are the times TrueScribe reaches the network.
Free forever
Core transcription and review without file or time limits.
US$59 one-time
One licence for two devices. Applicable taxes are handled at checkout.
No. Transcription, speaker detection, translation, editing and search run locally. TrueScribe uses the network to download models, to fetch a link you asked it to transcribe, to check your licence, and to report errors.
TrueScribe uses Whisper models, which cover a wide range of languages. It can detect the spoken language automatically or use the language you select.
No. TrueScribe runs on the CPU and can use a Vulkan runtime when your hardware supports it. The suggested model is matched to the machine it runs on, and faster hardware mainly means faster results.
Yes, once the models you use have been downloaded. The first transcription needs a connection to fetch the selected model.
No. Pro is a one-time purchase.
Windows and Linux, distributed through the Microsoft Store and Flathub.
Install TrueScribe, open a file, and read what is in it.
