Using LSTMs with data with multiple channels

Hi there!

I have developed a network composed of a few stacked LSTMs that analyses waveforms and tells me whether they contain a signal or just noise.
If I wanted to input a waveform as seen by three detectors(channels) (i.e. three different waveforms going into the network at the same time) how would i do this?
Would using convolutional LSTMs help with this?

If your unbatched input is 2d, i.e. (time,3), a module can either cut it as (1,3) pieces - that’s RNNs (that are memory accumulation models) or MLP modules (function of 3d vectors), or as (kernel_size,3) - that’s conv1d, appropriate for detecting temporal patterns. ConvLSTM needs extra spatial dimension I think.