I'm not clear how is the input features are stacked in wakeword applications. Taking https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/43969.pdf as an example, the author mentioned that

First, in the feature extraction module, 40 dimensional log-mel filterbank features are computed every 25ms with a 10ms frame shift. Next, at every frame, we stack 23 frames to the left and 8 frames to the right, and input this into the DNN.

So, given a one second audio, sample rate: 16000, frame length: 25ms, frame shift: 10ms and mel bins: 40. We can have a feature with the shape of [98, 40], where 98 is the number of frames.

How can we stack this feature? Thanks


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.