Unsupervised speaker adaptation for telephone call transcription

Wallace, Roy; Thambiratnam, Kit; Seide, Frank

doi:10.1109/icassp.2009.4960603

Cited by 9 publications

(7 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The conventional adaptation techniques can be handled using two types of learning approach: supervised and unsupervised (Matsui and Furui, 1998;Wallace et al, 2009). In supervised adaptation, prior knowledge of adaptation data, such as speaker information or manually defined labels, is required.…”

Section: Conventional Adaptation Approachesmentioning

confidence: 99%

Unsupervised rapid speaker adaptation based on selective eigenvoice merging for user-specific voice interaction

Choi

Park

2015

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

Section: Conventional Adaptation Approachesmentioning

confidence: 99%

Unsupervised rapid speaker adaptation based on selective eigenvoice merging for user-specific voice interaction

Choi

Park

2015

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

“…Transfer learning without any labeled data from the target domain is referred to as unsupervised transfer learning. Motivated by the success of unsupervised transfer learning for speaker adaptation (Chen et al, 2011;Wallace et al, 2009) and spoken document summarization (Lee et al, 2013), we further investigate whether unsupervised transfer learning is feasible for QA.…”

Section: Transfer Learning For Qamentioning

confidence: 99%

Supervised and Unsupervised Transfer Learning for Question Answering

Chung¹,

Lee²,

Glass³

2018

Proceedings of the 2018 Conference of the North American Chapter Of the Association for Computational Linguistics: Hu

View full text Add to dashboard Cite

Although transfer learning has been shown to be successful for tasks like object and speech recognition, its applicability to question answering (QA) has yet to be well-studied. In this paper, we conduct extensive experiments to investigate the transferability of knowledge learned from a source QA dataset to a target dataset using two QA models. The performance of both models on a TOEFL listening comprehension test (Tseng et al., 2016) and MCTest (Richardson et al., 2013) is significantly improved via a simple transfer learning technique from MovieQA (Tapaswi et al., 2016). In particular, one of the models achieves the state-of-the-art on all target datasets; for the TOEFL listening comprehension test, it outperforms the previous best model by 7%. Finally, we show that transfer learning is helpful even in unsupervised scenarios when correct answers for target QA dataset examples are not available.

show abstract

“…To date, direct matching of long phonetic sequences has been demonstrated only in rather restricted experimental settings (notably, in English, and with a "topically clumpy" test collection in which people talk about a limited range of predefined topics). Considerable work needs to be done to determine whether the salient phonetic differentiations in specific languages are adequately captured by existing phonetic recognizers, the degree to which we can leverage long-term use of the same mobile phone by the same user to perform unsupervised adaptation to the way a specific searcher (or content producer) speaks [30], whether "query-talk" will match the content to be searched adequately well, and how well these techniques will work with richer and more representative collections.…”

Section: Designmentioning

confidence: 99%

Query by babbling

Oard

2012

Proceedings of the First Workshop on Information and Knowledge Management for Developing Region

View full text Add to dashboard Cite

The Spoken Web, an interconnected collection of spoken content accessed through audio-only mobile phones, holds the promise of transforming information access for users in developing regions. The scale of the Spoken Web is, however, limited because current speech retrieval technology is only affordably deployable for a handful of languages. This paper proposes rethinking the conventional keyword query paradigm to instead develop systems that support a longer, richer, and more fluid interaction style that is better suited to both the affordances of spoken interaction and to the limitations of current speech technology.

show abstract

Unsupervised speaker adaptation for telephone call transcription

Cited by 9 publications

References 8 publications

Unsupervised rapid speaker adaptation based on selective eigenvoice merging for user-specific voice interaction

Unsupervised rapid speaker adaptation based on selective eigenvoice merging for user-specific voice interaction

Supervised and Unsupervised Transfer Learning for Question Answering

Query by babbling

Contact Info

Product

Resources

About