Perform unsupervised acoustic word segmentation using both unigram and (partially) bigram language models. The latter is only partially implemented; specifically, only bigram cluster assignments are sampled, and only using a maximum likelihood bigram language model.
If you use this code, please cite one of the references below (or all of them if you are kind). The code here relies on prior feature extraction and a number of preprocessed files for your corpus which is not illustrated here. See bucktsong_segmentalist for a complete recipe using this code.
Run make to build the Cython components. Run make test to run unit tests.
An IPython notebook example of clustering is given in
examples/clustering_examples.ipynb. This is just to illustrate some of the
differences between FBGMM and k-means clustering, and no segmentation is
performed on the generated toy data.
- H. Kamper, A. Jansen, and S. J. Goldwater, "A segmental framework for fully-unsupervised large-vocabulary speech recognition," arXiv preprint arXiv:1606.06950, 2016.
- H. Kamper, A. Jansen, and S. J. Goldwater, "Unsupervised word segmentation and lexicon discovery using acoustic word embeddings," IEEE Trans. Audio, Speech, Language Process., vol. 24, no. 4, pp. 669-679, 2016.
- H. Kamper, A. Jansen, and S. J. Goldwater, "Fully unsupervised small-vocabulary speech recognition using a segmental Bayesian model," in Proc. Interspeech, 2015.