matiaslindgren

Results 15 issues of matiaslindgren

Add hands-on tutorial with thorough command line examples to show where all stuff on the tutorial page is coming from

Code review of all Rust-code and reduce unnecessary syntax noise

Gather possible issues from reddit and hackernews comments etc

Replace scipy-filter and resampling with tf ops

[filter](https://github.com/py-lidbox/lidbox/blob/b13aca44c7bbb66d7a968fa040bdf8f1d9b3d553/lidbox/features/audio.py#L64) [resample](https://github.com/py-lidbox/lidbox/blob/b13aca44c7bbb66d7a968fa040bdf8f1d9b3d553/lidbox/features/audio.py#L37)

Code review of the feature extraction package

https://github.com/py-lidbox/lidbox/tree/master/lidbox/features Especially the correctness of DSP-related functions.

Add generic Kaldi-like metadata loader

utt2path, utt2label etc -> pandas.DataFrame

Cleanup embed-package

Rewrite end-to-end builder or decide not to use config files at all

Defining an end-to-end pipeline in yaml adds an unnecessary layer of complexity. Perhaps a single example pipeline could be supported but any customization is easier to do with a custom...

Inspect if model2function ops are placed on a device during function creation or when the function is called

https://github.com/py-lidbox/lidbox/blob/49c27a00eb4f02d085a4e470e08a00c427a042e0/lidbox/util.py#L101-L106

Support for ragged batches during training

For example, the x-vector architecture should be trained on arbitrary length input. Without ragged batches, this limits the batch size to 1. By supporting ragged batches, we could train with...

optimization