Fully unsupervised small-vocabulary speech recognition using a segmental Bayesian model

By Herman Kamper, Aren Jansen, Sharon Goldwater. 2015

42 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Probabilistic & Bayesian Methods, Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. James R. Glass
2. David F. Harwath
3. Herman Kamper
4. Karen Livescu
5. Wei-Ning Hsu
6. Sharon Goldwater
7. Aren Jansen
8. Yu-An Chung

In Probabilistic & Bayesian Methods

1. David F. Harwath
2. James R. Glass
3. Wei-Ning Hsu
4. Karen Livescu
5. Aren Jansen
6. Herman Kamper
7. Jennifer Drexler
8. Sharon Goldwater

In Speech & Audio

1. James R. Glass
2. David F. Harwath
3. Herman Kamper
4. Karen Livescu
5. Wei-Ning Hsu
6. Sharon Goldwater
7. Shane Settle
8. Aren Jansen

In Signal Processing

1. James R. Glass
2. Karen Livescu
3. Herman Kamper
4. David F. Harwath
5. Wei-Ning Hsu
6. Sharon Goldwater
7. Aren Jansen
8. Yu-An Chung

Citations in the corpus

• Learning Hierarchical Discrete Linguistic Units from Visually-Grounded Speech (2019, methodology): “…models have been successfully used to learn powerful speech representations using weakly or unsupervised objectives (Thiolliere et al., 2015; Kamper et al., 2015; Hsu et al., 2017a;b; Hsu & Glass, 2018; Holzenberger et al., 2018; Milde & Biemann, 2018; van den Oord et al., 2018; Chung et…”
• Unsupervised Speech Recognition via Segmental Empirical Output Distribution Matching (2018, methodology): “Another line of research uses a simpler segmentation method as an initialization, and jointly trains the segmenting and acoustic models for phonemes or words (Kamper et al., 2015; Glass, 2003; Siu et al., 2014; Lee and Glass, 2012).”
• Unsupervised Word Segmentation and Lexicon Discovery Using Acoustic Word Embeddings (2016, methodology): “The hyperparameters we used in [14] led to far less consistent performance over multiple sampling runs: WER standard deviations were in the order of 9% absolute, compared to the deviations of less than 1% that we obtain in Section IV-C.”
• Deep convolutional acoustic word embeddings using word-pair side information (2015, methodology): “Such embeddings have subsequently been used for keyword search [14] and unsupervised term discovery [17].”
• Discriminative acoustic word embeddings: Tecurrent neural network-based approaches (2016, methodology): “This representation was subsequently used in several applications: a segmental approach for query-by-example search [13], lexical clustering [19], and unsupervised speech recognition [20].”
• Analysis of Audio-Visual Features for Unsupervised Speech Recognition (2017, methodology): “Table 5: Unsupervised recognition results on the Buckeye corpus, using the system described in [16].”