anteju

Results 5 comments of anteju

[Fix] Use number of channels when calculating BAN

@boeddeker, possibly of interest -- this seems to apply to `pb_*` repos as well.

[Fix] Use number of channels when calculating BAN

@boeddeker & @desh2608, I just wanted to let you know, up to you whether to include it or not. Since you mentioned CHiME: processed audio is saved to fixed point...

Audio model collection

> LGTM overall. We could also rename `https://github.com/anteju/NeMo/blob/pr/audio-collection/examples/audio/speech_enhancement.py` to `https://github.com/anteju/NeMo/blob/pr/audio-collection/examples/audio/audio_to_audio_train.py` to align with rest of the filenames in the examples folder Done.

Enormous number of `.nemo` checkpoints produced in training

@mikolajblaz, maybe the above is related to #9015?

Question about the settings in speech_data_simulator

@tango4j, could you check the above issue with `num_speakers` and `sentence_length_params`?