Fast and Robust Features for Prosodic Classification?

Buckow, Jan; Warnke, Volker; Huber, Richard; Batliner, Anton; Nöth, Elmar; Niemann, Heinrich

doi:10.1007/3-540-48239-3_35

Cited by 7 publications

(8 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We developed a database of 270 prosodic features describing pause, pitch, duration, and energy information in the vicinity of each word boundary, inspired by [2,17]. Features were extracted directly from the automatically aligned speech signal, so that no hand-labeling of prosody (such as ToBI) was necessary in model training.…”

Section: Prosodic Featuresmentioning

confidence: 99%

Using Prosody for Automatic Sentence Segmentation of Multi-party Meetings

Kolář

Shriberg

Liu

2006

Text, Speech and Dialogue

View full text Add to dashboard Cite

Abstract. We explore the use of prosodic features beyond pauses, including duration, pitch, and energy features, for automatic sentence segmentation of ICSI meeting data. We examine two different approaches to boundary classification: score-level combination of independent language and prosodic models using HMMs, and feature-level combination of models using a boosting-based method (BoosTexter). We report classification results for reference word transcripts as well as for transcripts from a state-of-the-art automatic speech recognizer (ASR). We also compare results using the lexical model plus a pause-only prosody model, versus results using additional prosodic features. Results show that (1) information from pauses is important, including pause duration both at the boundary and at the previous and following word boundaries; (2) adding duration, pitch, and energy features yields significant improvement over pause alone; (3) the integrated boosting-based model performs better than the HMM for ASR conditions; (4) training the boosting-based model on recognized words yields further improvement.

show abstract

Section: Prosodic Featuresmentioning

confidence: 99%

Using Prosody for Automatic Sentence Segmentation of Multi-party Meetings

Kolář

Shriberg

Liu

2006

Text, Speech and Dialogue

View full text Add to dashboard Cite

show abstract

“…We originally developed a database of 270 prosodic features (inspired by [3,17]) that capture pause, pitch, duration, and energy information associated with each word boundary. Features were extracted directly from the automatically aligned speech signal.…”

Section: Prosodic Featuresmentioning

confidence: 99%

On speaker-specific prosodic models for automatic dialog act segmentation of multi-party meetings

Kolář¹,

Shriberg²,

Liu³

2006

Interspeech 2006

View full text Add to dashboard Cite

“…One practical problem with using segmental durations in the feature set, for developers using third-party (off-the-shelf) speech recognition software, is that commercial systems typically do not provide phonetic time alignments with the word recognition output, though word times may be available. In [7], a substitute set of features is proposed that does not require phone time alignments but instead uses word durations normalized by summed phoneme average duration statistics. While they report good results for accent and boundary detection, the approach obscures cues known to occur at the syllable level.…”

Section: Introductionmentioning

confidence: 99%

Non-segmental duration feature extraction for prosodic classification

et al. 2008

View full text Add to dashboard Cite

This paper presents a set of novel duration features for detecting pitch accent and phrase boundaries, which depend on articulatory timing rather than segmental duration information. The features are computed from the detected syllable nuclei and boundaries, using peaks and valleys in an energy contour but also leveraging information from a simple HMM phone manner class recognizer to increase recall. In experiments on the hand-segmented TIMIT corpus, we obtain greater than 90% Fmeasure for vowel detection. In prosody detection experiments on the BU Radio News corpus, comparing to a segmental feature baseline, we obtain similar performance for pitch accent detection and slightly worse boundary detection from the new features without the need for phonetic alignments.

show abstract

Fast and Robust Features for Prosodic Classification?

Cited by 7 publications

References 4 publications

Using Prosody for Automatic Sentence Segmentation of Multi-party Meetings

Using Prosody for Automatic Sentence Segmentation of Multi-party Meetings

On speaker-specific prosodic models for automatic dialog act segmentation of multi-party meetings

Non-segmental duration feature extraction for prosodic classification

Contact Info

Product

Resources

About