Non-traditional prosodic features for automated phrase break prediction

Brierley, Claire; Atwell, Eric

doi:10.1093/llc/fqr023

Cited by 4 publications

(3 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Especially in the context of speech synthesis research, many attempts to predict intonation group boundaries have also drawn on grammatical information and punctuation. More recent work has additionally proposed incorporating other non-traditional cues -for instance, recognizing a link between prosodic boundaries and the presence of words that include dipththongs or triphthongs (Brierley & Atwell, 2011).…”

Section: Aimsmentioning

confidence: 99%

Units and Constituency in Prosodic Analysis: A Quantitative Assessment

Wilson

2016

Journal of Quantitative Linguistics

View full text Add to dashboard Cite

Drawing on methods from quantitative linguistics, this paper tests the hypothesis that the intonation unit is a valid language construct whose immediate constituent is the foot (and whose own immediate constituent is the syllable). If the hypothesis is true, then the lengths of intonation units, measured in feet, should abide by a regular and parsimonious discrete probability distribution, and the immediate constituency relationship between feet and intonation units should be further demonstrable by successfully fitting the Menzerath-Altmann equation with a negative exponent. However, out of sixteen texts from the Aix-MARSEC database, only six share a common probability distribution and only eight exhibit a tolerable fit of the Menzerath-Altmann equation. A failure rate of ≥ 50% in both cases casts doubt on the validity of the hypothesis.

show abstract

Section: Aimsmentioning

confidence: 99%

Units and Constituency in Prosodic Analysis: A Quantitative Assessment

Wilson

2016

Journal of Quantitative Linguistics

View full text Add to dashboard Cite

show abstract

“…Intonational phrase structure is a well studied problem in linguistics by phoneticians, psycholinguists and syntacticians alike [3][4][5][6]. The results of such explorations has been applied to text-to-speech synthesis in many instances [7][8][9].…”

Section: Relation To Prior Workmentioning

confidence: 99%

Intonational phrase break prediction for text-to-speech synthesis using dependency relations

Mishra

Kim

Bangalore

2015

2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Intonational phrase (IP) break prediction is an important aspect of front-end analysis in a text-to-speech system. Standard approaches for intonational phrase break prediction rely on the use of linguistic rules or more recently, lexicalized data-driven models. Linguistic rules are not robust while data-driven models based on lexical identity do not generalize across domains. To overcome these challenges, in this paper, we explore the use of syntactic features to predict intonational phrase breaks. On a test set of over 40 thousand words, while a lexically driven IP break prediction model yields an F-score of 0.82, a non-lexicalized model that uses part-of-speech tags and dependency relations achieves an F-score of 0.81 with added feature of being more portable across domains. In this work, we also examine the effect of contextual information on prediction performance. Our evaluation shows that using a three-token left context in a POS-tag based model results in only a 2% drop in recall compared to a model that uses both a left and right context, which suggests the viability of using such a model for incremental text-to-speech system.

show abstract

“…The prediction of pauses or prosodic breaks from text is a wide-studied topic in the field of TTS research. Methods range from rule based approaches [1,2,3] to a variety of machine learning methods like decision trees [4], HMMs [5,6], Conditional Random Fields (CRFs) [7], Bidirectional Recurrent Neural Networks (BiRNNs) [8], and also combinations of methods [9,10,11,12,13]. There are many links into other fields such as linguistics, syntax, psychology and automatic speech recognition which include transcription systems for intonational phrase structure (i.e.…”

Section: Introductionmentioning

confidence: 99%

Pause Prediction from Text for Speech Synthesis with User-Definable Pause Insertion Likelihood Threshold

Braunschweiler

Maia

2016

Interspeech 2016

View full text Add to dashboard Cite

Predicting the location of pauses from text is an important aspect for speech synthesizers. The accuracy of pause prediction can significantly influence both naturalness and intelligibility. Pauses which help listeners to better parse the synthesized speech into meaningful units are deemed to increase naturalness and intelligibility ratings, while pauses in unexpected or incorrect locations can reduce these ratings and cause confusion. This paper presents a multi-stage pause prediction approach including first prosodic chunk prediction, followed by a feature scoring algorithm and finally a pause sequence evaluation module. Preference tests showed that the new method outperformed a pauses-at-punctuation baseline while not yet matching human performance. In addition, the approach includes two more functionalities: (1) a user-specifiable pause insertion rate and (2) multiple output formats in the form of binary pauses, multi-level pauses or as a score reflecting pause strength.

show abstract

Non-traditional prosodic features for automated phrase break prediction

Cited by 4 publications

References 2 publications

Units and Constituency in Prosodic Analysis: A Quantitative Assessment

Units and Constituency in Prosodic Analysis: A Quantitative Assessment

Intonational phrase break prediction for text-to-speech synthesis using dependency relations

Pause Prediction from Text for Speech Synthesis with User-Definable Pause Insertion Likelihood Threshold

Contact Info

Product

Resources

About