Universal attribute characterization of spoken languages for automatic spoken language recognition

Siniscalchi, Sabato Marco; Reed, Jeremy; Svendsen, Torbjørn; Lee, Chin‐Hui

doi:10.1016/j.csl.2012.05.001

Cited by 51 publications

(43 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, adding context information allowed substantially better results. So far, we have only used manner of articulation features, yet place of articulation can further enhance accent recognition performance, as shown in [16]. As a future work, experiments on English foreign accent recognition will be carried out.…”

Section: Resultsmentioning

confidence: 99%

“…Furthermore, data-sharing across languages at the acoustic phonetic attribute level is naturally facilitated by using these attributes, so more reliable language-independent acoustic parameter estimation can be anticipated [21]. In [16], it was also shown that these attributes can be used to compactly characterize any spoken language along the same lines as in the automatic speech attribute transcription (ASAT) paradigm for automatic speech recognition (ASR) [20]. Therefore, we believe that it can also be useful to characterize speaker accent.…”

Section: Choice and Extraction Of Attribute Featuresmentioning

confidence: 99%

“…However, knowledge based modeling, such as phonotactic features, are known to be linguistically and phonetically relevant [5]. However, the front-end of the phonotactic system needs a tokenizer that will turn the utterance into a sequence of "phonetic letters" [15,16]. An ad-hoc approach is to use a phone recognizer developed for one language, such as Hungarian, and apply it to all phonotactic recognition tasks [17].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Introducing attribute features to foreign accent recognition

Behravan

Siniscalchi

Kinnunen

et al. 2014

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

We propose a hybrid approach to foreign accent recognition combining both phonotactic and spectral based systems by treating the problem as a spoken language recognition task. We extract speech attribute features that represent speech and acoustic cues reflecting foreign accents of a speaker to obtain feature streams that are modeled with the i-vector methodology. Testing on the Finnish Language Proficiency exam corpus, we find our proposed technique to achieve a significant performance improvement over the state-of-the-art systems using only spectral based features.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Choice and Extraction Of Attribute Featuresmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Introducing attribute features to foreign accent recognition

Behravan

Siniscalchi

Kinnunen

et al. 2014

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…As mentioned above, training a LID system in a semi-supervised setting is done with collections of labeled and unlabeled data. Note that by "training data" we do not mean data useful for learning how to generate the articulatory features; in fact, the approach mentioned in [6] does not require the articulatory attribute generator to be trained on the languages that it will eventually be applied on.…”

Section: Introductionmentioning

confidence: 99%

“…The attribute-based approach proposed by [6] (which we make use of in this paper) explores the universal acoustic phonetic features of speech with state-of-the-art performance results. Specifically, using HMMs and a language-dependent to language-independent mapping procedure, raw speech is converted into "manner" and "place" of articulation attributes which are subsequently used to train HMM-based and SVMbased identification systems.…”

Section: Introductionmentioning

confidence: 99%

Bootstrapping a spoken language identification system using unsupervised integrated sensing and processing decision trees

Huang

Karakos

Coppersmith

et al. 2011

2011 IEEE Workshop on Automatic Speech Recognition &Amp; Understanding

Self Cite

View full text Add to dashboard Cite

Abstract-In many inference and learning tasks, collecting large amounts of labeled training data is time consuming and expensive, and oftentimes impractical. Thus, being able to efficiently use small amounts of labeled data with an abundance of unlabeled data-the topic of semi-supervised learning (SSL) [1]-has garnered much attention. In this paper, we look at the problem of choosing these small amounts of labeled data, the first step in a bootstrapping paradigm. Contrary to traditional active learning where an initial trained model is employed to select the unlabeled data points which would be most informative if labeled, our selection has to be done in an unsupervised way, as we do not even have labeled data to train an initial model.We propose using unsupervised clustering algorithms, in particular integrated sensing and processing decision trees (ISPDTs) [2], to select small amounts of data to label and subsequently use in SSL (e.g. transductive SVMs). In a language identification task on the CallFriend 1 and 2003 NIST Language Recognition Evaluation corpora [3], we demonstrate that the proposed method results in significantly improved performance over random selection of equivalently sized training data.

show abstract