Transcribe the audio of any tab on your own computer and save it to a text file. Works offline; audio never leaves your device.
Users
45
store count
Rating
0.0 / 5
unchanged in 30 days
Ratings
0
total ratings
Store rank
#114K
#95K among extensions
Version
1.0.0
updated 7 Oct 2026
About Local Transcript (Speech-to-Text)
Transcribe lectures, meetings, podcasts, and videos directly in your browser — using speech recognition that runs entirely on your own computer. Nothing is sent to a server. Audio never leaves your device.
How it works
1. Click the Local Transcript icon to open the side panel on any tab.
2. Choose a text file to save the transcript to.
3. Press Start — the extension begins transcribing in real time.
4. Press Stop when you're done. The transcript is already saved to your file.
What makes Local Transcript different
✔ Fully offline — the Whisper speech model (whisper-base.en) ships inside the extension. No internet connection required after installation.
✔ Private by design — audio is captured, transcribed, and immediately discarded on your device. No account, login, or cloud service of any kind.
✔ Saves as you go — the transcript is written to a plain text file in real time, chunk by chunk. Even if you close the browser mid-session, everything transcribed so far is kept.
✔ Works on any tab — lectures on Coursera, Teams meetings, YouTube videos, podcasts, Zoom calls. Any tab that plays audio can be transcribed.
✔ Keeps recording with the panel closed — the side panel is just for display; closing it doesn't interrupt the recording.
✔ Stops by itself — the recording ends automatically when the video or audio finishes, or after a period of silence, so you don't have to watch it.
✔ Optional timestamps — each paragraph can be prefixed with its recording time (e.g. [1:05]), so you can find specific parts of the original video later.
✔ Uses your GPU when available — inference runs on WebGPU if your hardware supports it, falling back to WebAssembly otherwise.
✔ No subscription — Local Transcript is free to use.
Accuracy
Local Transcript uses OpenAI's whisper-base.en model. It handles clear speech well and is particularly suited to spoken English in lectures, podcasts, and presentations. Like all speech-to-text systems, accuracy varies with audio quality, accent, and background noise.
Privacy
See the full privacy policy for details. The short version: nothing leaves your computer. Ever.