Language-independent and language-adaptive acoustic modeling for speech recognition

By Tanja Schultz, A. Waibel. 2001

459 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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. Dong Yu
2. Jinyu Li
3. Li Deng
4. Yoshua Bengio
5. James R. Glass
6. Tara N. Sainath
7. Shinji Watanabe
8. Graham Neubig

In Speech & Audio

1. Jinyu Li
2. Dong Yu
3. Shinji Watanabe
4. Li Deng
5. S. Renals
6. Tara N. Sainath
7. Wei-Ning Hsu
8. Pawel Swietojanski

In Signal Processing

1. Dong Yu
2. Jinyu Li
3. Li Deng
4. S. Renals
5. Yifan Gong
6. Tara N. Sainath
7. Shinji Watanabe
8. Pawel Swietojanski

Citations in the corpus

• Multilingual Speech Recognition with a Single End-to-End Model (2017, background): “Some of the multilingual AMs require a common phone set [3, 4, 13] while others share some of the acoustic model parameters [9–11, 15].”
• Unsupervised Pretraining Transfers Well Across Languages (2020, background): “A common way to improving speech recognition in low-resource languages is to train multilingual speech recognition with shared components [8, 9, 2].”
• A Survey of Cross-lingual Word Embedding Models (2017, background): “Language-independent representations have been proposed for decades, many of which rely on abstract linguistic labels instead of lexical features (Aone & McKee, 1993; Schultz & Waibel, 2001).”
• Multilingual deep neural network based acoustic modeling for rapid language adaptation (2014, methodology): “To train a multilingual acoustic model, there are several possible ways [13]: on a merged universal phoneset based on the international phonetic alphabet (IPA) chart, i.”
• Speech Recognition Using Deep Neural Networks: A Systematic Review (2019, background): “The applications of automatic language recognition evidently appear in spoken language translation [27], multilingual speech recognition [28], and spoken document retrieval [29].”
• Sequence-Based Multi-Lingual Low Resource Speech Recognition (2018, background): “Authors in [10] introduce the use of a shared phone set to build HMM based language independent acoustic models and show the adaptation of pre-existing models towards a new language.”
• Current trends in multilingual speech processing (2011, methodology): “The approach of knowledge-based merging followed by data-driven clustering for context-dependent phone models was further investigated in [Schultz and Waibel, 2001].”
• Cross-language knowledge transfer using multilingual deep neural network with shared hidden layers (2013, methodology): “, the IPA universal phone set [15][16]) or by data-driven clustering [15][16][17][18].”
• Multilingual training of deep neural networks (2013, methodology): “Previous approaches in this direction have included: constructing a universal phone set [1, 2, 3]; modeling a set of universal speech attributes, such as voicing, nasality and frication [4]; and mapping between phones of different languages using some automated method that relies on some predefined…”
• Building a great multi-lingual teacher with sparsely-gated mixture of experts for speech recognition (2021, background): “There has been a great deal of attention to multi-lingual automatic speech recognition (ASR) that can recognize multiple languages [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20].”
• Unsupervised cross-lingual knowledge transfer in DNN-based LVCSR (2012, methodology): “A number of techniques for cross-lingual acoustic modelling have been published including the use of: global phone sets [1, 2]; multilingual posterior features in tandem [3, 4] and Kullback-Liebler hidden Markov model (KL-HMM) systems [5]; subspace Gaussian mixture models (SGMMs) with a shared…”
• Automatic speech recognition for under-resourced languages: A survey (2014, methodology): “…approaches have been employed, ranging from pooling data across languages [Van Heerden, 2010], through bootstrapping from source-model alignments [Schultz, 2001; Le, 2009], to phone mapping for recognition with the source models [Chan, 2012], possibly after some maximum a-posteriori (MAP)…”
• Universal Phone Recognition with a Multilingual Allophone System (2020, background): “The first group, shared phoneme models, creates a shared phoneme inventory of all phonemes from all training languages [3, 4, 5, 6, 7, 10].”
• Multilingually trained bottleneck features in spoken language recognition (2017, methodology): “In Schultz and Waibel (2001), International Phonetic Alphabet (IPA) was used to create a cross-lingual phoneme set by unifying the phoneme sets of different languages from this database.”
• A study on multilingual acoustic modeling for large vocabulary ASR (2009, background): “One potential solution is to explore shared acoustic phonetic structures among different languages to build a large set of acoustic models (e.g., [1][2][3][5][7][8][9]) that characterize all the phone units needed in order to cover all the spoken languages being considered.”