System information

• Have I written custom code (as opposed to using a stock example script
  provided in TensorFlow):No
• TensorFlow version (use command below): 2.12.0
• Python version: 3.10.*

Describe the feature and the current behavior/state.

The current version of TensorFlow's tf.signal module provides extensive support for various Fourier Transform functions such as fft() and rfft(). However, it does not include helper functions like fftfreq() and rfftfreq() available in other libraries, such as NumPy and PyTorch. These functions are used to compute the discrete Fourier Transform sample frequencies for a signal of a given size, which is a common requirement in many signal-processing tasks.

Currently, to achieve similar functionality, users have to define custom functions or import existing functions from other libraries, such as SciPy. This process requires converting TensorFlow tensors to and from the other library's format, which may not always be efficient or convenient.

Who will benefit from this feature?

Users who are working on signal processing tasks using TensorFlow will benefit from this feature as they won't have to switch to other libraries (like NumPy or PyTorch) to compute the Fourier Transform frequencies. This will make their code more consistent and potentially more efficient.

Additional Info.

Adding these functions will make the tf.signal module more complete and competitive with other libraries' offerings in terms of signal processing capabilities. It'll also make TensorFlow more user-friendly for those who are accustomed to these functions in other libraries.