There is already a "whisper" node, even with two strategies (ONNX and GGML), but their overall performance on a CPU(!) is dramatically slow. Either get it running on GPU or think about alternatives (use projects like https://speaches.ai or VOSK https://alphacephei.com/vosk/).