whisper.cpp

3989b29a - examples : fix n_gpu_layers usage in talk-llama (#1441)