Support new onnx signal operators

[hwangdeyu]

There are several signal operators added in ONNX 1.12. Tracking issue for it.

• DFT ONNX: https://github.com/onnx/onnx/blob/main/docs/Operators.md#DFT TensorFlow: https://www.tensorflow.org/api_docs/python/tf/signal/fft https://www.tensorflow.org/api_docs/python/tf/signal/ifft
• STFT ONNX: https://github.com/onnx/onnx/blob/main/docs/Operators.md#STFT TensorFlow: https://www.tensorflow.org/api_docs/python/tf/signal/stft
• HannWindow ONNX: https://github.com/onnx/onnx/blob/main/docs/Operators.md#HannWindow TensorFlow: https://www.tensorflow.org/api_docs/python/tf/signal/hann_window
• HammingWindow ONNX: https://github.com/onnx/onnx/blob/main/docs/Operators.md#HammingWindow TensorFlow: https://www.tensorflow.org/api_docs/python/tf/signal/hamming_window
• MelSpectrogram ONNX: https://github.com/onnx/onnx/blob/main/docs/Operators.md#MelWeightMatrix TensorFlow: https://www.tensorflow.org/api_docs/python/tf/signal/linear_to_mel_weight_matrix

[Yannick947]

Is there anyone working on this issue? Tensorflow models using these operations cannot be exported properly without using this hacky workaround that requires to install yet another library..

The opset is already implemented in plain onnx, so there is nothing blocking from implementing this feature for tf2onnx?

[Yilmazzn]

Ran into the same problem. It would be awesome to see this feature implemented soon