Closed qpjaada closed 1 year ago
We had an internal discussion about dither in the validation pipeline, and the conclusion was, that
dither is just small noise added to even out quantization artifacts. And it should not change WER. https://kaldi-asr.org/doc/faq.html#indeterminacy_in_feature_extraction
so this is not considered as an augmentation and represents industry use.
A non-zero value of
dither_coeff
is specified for feature extraction during validation here: https://github.com/mlcommons/training/blob/master/rnn_speech_recognition/pytorch/configs/baseline_v3-1023sp.yaml#L38This leads to non-deterministic results during validation because of randomness here: https://github.com/mlcommons/training/blob/master/rnn_speech_recognition/pytorch/common/data/dali/pipeline.py#L197
Seems like for validation-pipeline, dithering should be disabled.