A summary of the 2012 JHU CLSP workshop on zero resource speech technologies and models of early language acquisition
By Aren Jansen, Emmanuel Dupoux, Sharon Goldwater, Mark Johnson, Sanjeev Khudanpur, Kenneth Ward Church, Naomi H Feldman, Hynek Hermansky, Florian Metze, Richard C. Rose and 17 more. 2013
113 citations recorded by the source corpus.
We summarize the accomplishments of a multi-disciplinary workshop exploring the computational and scientific issues surrounding zero resource (unsupervised) speech technologies and related models of early language acquisition. Centered around the tasks of phonetic and lexical discovery, we consider unified evaluation metrics, present two new approaches for improving speaker independence in the absence of supervision, and evaluate the application of Bayesian word segmentation algorithms to automatic subword unit tokenizations. Finally, we present two strategies for integrating zero resource techniques into supervised settings, demonstrating the potential of unsupervised methods to improve mainstream technologies.
Subfields: Artificial Intelligence, Developmental and Educational Psychology, Experimental and Cognitive Psychology, Speech & Audio, Linguistics
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
- James R. Glass
- Wei-Ning Hsu
- Emmanuel Dupoux
- Karen Livescu
- Hung-yi Lee
- Herman Kamper
- Gabriel Synnaeve
- Yoshua Bengio
In Developmental and Educational Psychology
- James R. Glass
- Herman Kamper
- Karen Livescu
- Leanne Nortje
- Timothy J. O'Donnell
- Geoffry Zweig
- Hung-yi Lee
- Emmanuel Dupoux
In Experimental and Cognitive Psychology
- Emmanuel Dupoux
- Alexei Baevski
- Xuan-Nga Cao
- Gabriel Synnaeve
- Neil Zeghidour
- Leanne Nortje
- Timothy J. O'Donnell
- James R. Glass
In Speech & Audio
- James R. Glass
- Emmanuel Dupoux
- Herman Kamper
- Wei-Ning Hsu
- Karen Livescu
- Hung-yi Lee
- Abdel-rahman Mohamed
- David F. Harwath
In Linguistics
- James R. Glass
- Herman Kamper
- Karen Livescu
- Leanne Nortje
- Timothy J. O'Donnell
- Hung-yi Lee
- Emmanuel Dupoux
- Benjamin van Niekerk
Citations in the corpus
- Unsupervised acoustic unit discovery for speech synthesis using discrete latent-variable neural networks (2019, background): “A central problem is finding frame-level feature representations that capture meaningful linguistic contrasts, such as phonetic categories, while being invariant to factors such as a speaker’s identity, gender or accent [2–5].”
- A deep scattering spectrum — Deep Siamese network pipeline for unsupervised acoustic modeling (2016, background): “In the task of learning an acoustic model, unsupervised systems are on the rise [1], however the gap between these systems and the supervised ones remains considerable.”
- Audio Word2vec: Sequence-to-Sequence Autoencoding for Unsupervised Learning of Audio Segmentation and Representation (2019, background): “Word segmentation of speech is critical but challenging for zero-resource speech technology because word boundaries are usually not available for given speech utterances or corpora [3], [18]–[20].”
- Self-Supervised Speech Representation Learning: A Review (2022, unknown): “, transcriptions and other annotations [302].”
- Unsupervised Lexicon Discovery from Acoustic Input (2015, result): “Finally, note that even though the F1 scores for the word segmentation task are low, we find similar performance reported in Jansen et al. (2013).”
- Evaluating speech features with the minimal-pair ABX task (II): resistance to noise (2014, background): “These metrics are problematic in zeroresource/unsupervised settings (were only very limited amount of labeled data, if any, is available for training [1, 2]).”
- Evaluating speech features with the minimal-pair ABX task: analysis of the classical MFC/PLP pipeline (2013, background): “[10, 11, 12, 13, 14, 15, 16, 17, 18]), other metrics [19] as well as human performance could be investigated within our framework.”
- Cyclic Spectral Modeling for Unsupervised Unit Discovery into Voice Conversion with Excitation and Waveform Modeling (2020, background): “Recently, a lot of work have been carried out on the so-called zero resource settings of speech processing [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13].”
- Semantic Speech Retrieval With a Visually Grounded Model of Untranscribed Speech (2017, background): “Also motivated by the problem of modelling language acquisition, there is growing interest in the speech community in models that can learn directly from unlabelled speech [34], [35].”
- Segmental acoustic indexing for zero resource keyword search (2015, background): “As a result, the zero-resource setting, in which detailed annotations are unavailable and linguistic structure must be discovered without the aid of training data, has attracted attention both in the speech processing community [2] and among scientists interested in human language acquisition [3].”
- Effectiveness of Self-Supervised Pre-Training for ASR (2020, background): “The practical need for building ASR systems for new conditions with limited resources spurred a lot of work focused on unsupervised speech recognition and representation learning [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14], in addition to semi- and weakly-supervised learning techniques to reduce th”
- Large scale weakly and semi-supervised learning for low-resource video ASR (2020, background): “Reducing the amount of manual transcription for building speech recognition systems has been a constant research theme over the past three decades [40, 20, 9, 5, 41, 42] given the revolutionary increase and access to unlabeled data and computational resources.”
- ABX-Discriminability Measures and Applications. (Mesures de discriminabilité ABX et applications) (2016, methodology): “Yet our efforts to build artificial systems capable of the same achievement are only beginning, and state-of-the-art systems still have very poor performance [54].”