about
About MurmurScribe
MurmurScribe is a browser workspace for turning recordings into useful text. It is made by a small independent team.
Bring the model to the recording
A conversation can be useful long after it ends. A transcript makes it easier to find a phrase, prepare a quote, or create captions. We built MurmurScribe around local processing so those tasks do not require sending the recording to a transcription server.
The app uses OpenAI’s Whisper speech models through Transformers.js. Your browser downloads the chosen model and runs it on the device. The first run involves a download; later runs can reuse browser-cached files. Device memory and browser capabilities still determine what will work.
A draft you can inspect
Speech recognition is not a substitute for listening. The editor keeps timestamps close to the text so you can return to the recording, correct a name, and check a phrase before using it. Speaker labels are entered manually. We do not claim to identify the people in a conversation automatically.
Exports let you take the corrected draft into your own work. Plain text is useful for search and notes. A document supports later editing. Subtitle formats connect text to playback time. Choosing an export should follow the destination, not a promise that one format works everywhere.
Evidence with context
Language pages show results from the original Whisper paper. They explain which dataset and metric produced a score, and they distinguish those results from the quantized models running here. Comparison pages use dated official sources and acknowledge the features that other products do better.
We do not publish invented reviews, user counts or accuracy guarantees. A practical sample from your own recording is more useful than a broad marketing claim. When a source does not support a number, we leave it out and explain the uncertainty where it matters.
A clear product boundary
The current app is free. It has no account requirement or product minutes cap. A future Pro plan is described only as in development on the pricing page. Today’s tool stands on its own, with local storage controls and exports available to everyone using a supported browser.
A workspace built around the recording
