gaoxinglong/word2vec
Automatically exported from code.google.com/p/word2vec
Public repository discovered through GitHub real-time crawl.
Public repository record indexed from GitHub. Explore verified star velocity metrics, source code repositories, and curated developer tool directories across the GitHubRepo ecosystem.
Automatically exported from code.google.com/p/word2vec
My attempt to use a publicly available subset of RM1 speech database(MFCC feats from CMU and docs from LDC) to train and test Kaldi models. The existing egs/rm/s1 recipe from the Kaldi distribution was used as a template.
some useful algorithms