Convert kaldi feature extraction and nnet3 models into Tensorflow Lite models. Currently aimed at converting kaldi's x-vector models and diarization pipelines to tensorflow models.
Apache License 2.0
20
stars
3
forks
source link
refactor(lib/layers): dynamic input shape for `Framing` and more tests #6
Refactored the Framing layer to be able to handle dynamic input
shapes at inference time. The input to the model at build time can
have unknown shape; the shape must then be be set at inference time
using resize_input_tensor() and allocate_tensors().
Updated the unit tests for FilterBank and MFCC to also test
dynamic input shapes.
Updated docstrings in MFCC and fixed get_config().
Refactored the
Framing
layer to be able to handle dynamic input shapes at inference time. The input to the model at build time can have unknown shape; the shape must then be be set at inference time using resize_input_tensor() and allocate_tensors().Updated the unit tests for
FilterBank
andMFCC
to also test dynamic input shapes.Updated docstrings in
MFCC
and fixedget_config()
.