-
由 suleshahid 提交于
Second OP for the TFLM Signal library, Real-Valued Fast Fourier Transform. The RFFT OP provides three resolutions: `FLOAT, INT16, INT32` Similar usage as to previous Window OP: * `op_resolver.AddRfft()` (which will add all resolutions, and determine the type at runtime) * `op_resolver.AddRfftFloat()`, `op_resolver.AddRfftInt16()`, `op_resolver.AddRfftInt32()` for a specific resolution type. * or via python as can be seen in `fft_ops_test.py` 3 testing options are provided: * Micro(C++): bazel run signal/micro/kernels:fft_test * Tensorflow/Micro(Python): bazel run python/tflite_micro/signal:fft_ops_test * Makefile(C++): make -f tensorflow/lite/micro/tools/make/Makefile test_kernel_fft_test BUG=[287346710](http://b/287346710)
f22888af