This ReaScript provides real-time audio transcription in REAPER by integrating whisper.cpp's whisper-stream tool. The script automatically starts transcribing when Reaper is recording and adds the text as project markers on the timeline.
- REAPER: The script must be run inside the Reaper DAW. [7.40]
- Python for ReaScript: A configured Python 3 installation is required. [3.13]
- whisper.cpp: A compiled version of the
whisper.cppproject. [1ad258c] - A Whisper Model: A GGML-format model file (e.g.,
ggml-base.en.bin).
- Compile whisper.cpp with
-DWHISPER_SDL2=ONto allow for audio device input. - Run
./build/bin/whisper-streamin your command prompt to determine the ID of your intended capture device. This should match the input device you are using in Reaper. - Open the
whisper_stream_to_markers.pyscript and edit theWHISPER_DEVICE,WHISPER_EXECUTABLEandWHISPER_MODELvariables to match the details on your system.
Run the script from the Reaper Action List to open its console and begin monitoring. When you press record in Reaper, the script will automatically launch whisper-stream and start transcribing. Markers will appear on the timeline as you record.
When you stop recording, the transcription process will automatically stop and the script will terminate.
You can adjust the LENGTH_MS (audio chunk length) and STEP_MS (processing interval) values in the script's configuration section. These values determine the g_commit_interval, which controls how often markers are created. Larger values tend to increase the accuracy and reduce responsiveness.