Combining Vocal Tract Length Normalization With Hierarchical Linear Transformations

Saheer, Lakshmi Babu; Yamagishi, Junichi; Garner, Philip N.; Dines, John

doi:10.1109/jstsp.2013.2295554

Cited by 5 publications

(7 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, using a combination of adult and child training data in the DNN-based models resulted in improved results [24][25][26]. In particular, combining child speech with adult female speech has been proven to be advantageous [25,55]. As the length of an adult female vocal tract is closer to the length of a child's vocal tract in comparison to that of males, manipulating an adult female voice into a child's voice has often been more successful [25].…”

Section: Speech-synthesis Systemsmentioning

confidence: 99%

“…As the length of an adult female vocal tract is closer to the length of a child's vocal tract in comparison to that of males, manipulating an adult female voice into a child's voice has often been more successful [25]. In fact, the child voice that was adapted from an average-male-voice model experienced significantly larger distortion than that adapted from the average-female-voice model [55]. Similarly, a model trained on a male voice resulted in a less naturalistic child voice [47,56].…”

Section: Speech-synthesis Systemsmentioning

confidence: 99%

“…Alternatively, one could apply VTLN to the child's utterances during training and testing to make them more similar to the adults' speech [25]. Saheer et al's [55] study shows that combining VTLN with CSMAPLR resulted in an improved adaptation method for both HMM-based automatic speech recognition and text-to-speech. It was clear that when there was a limited amount of adaptation data available, even as little as one sentence, the VTLN yielded the best naturalness and intelligibility results.…”

Section: Speech-synthesis Systemsmentioning

confidence: 99%

“…Lastly, the clustered HMMs were re-estimated using speaker adaptive training, with regression classes for the normalisation being determined from the gender-mixed decision trees [20]. However, in contrast, gender-dependent average-voice models were used by [25,55] where improved performance, as compared to the baseline, was reported.…”

Section: Child-speech Datamentioning

confidence: 99%

See 3 more Smart Citations

A Situational Analysis of Current Speech-Synthesis Systems for Child Voices: A Scoping Review of Qualitative and Quantitative Evidence

et al. 2022

View full text Add to dashboard Cite

(1) Background: Speech synthesis has customarily focused on adult speech, but with the rapid development of speech-synthesis technology, it is now possible to create child voices with a limited amount of child-speech data. This scoping review summarises the evidence base related to developing synthesised speech for children. (2) Method: The included studies were those that were (1) published between 2006 and 2021 and (2) included child participants or voices of children aged between 2–16 years old. (3) Results: 58 studies were identified. They were discussed based on the languages used, the speech-synthesis systems and/or methods used, the speech data used, the intelligibility of the speech and the ages of the voices. Based on the reviewed studies, relative to adult-speech synthesis, developing child-speech synthesis is notably more challenging. Child speech often presents with acoustic variability and articulatory errors. To account for this, researchers have most often attempted to adapt adult-speech models, using a variety of different adaptation techniques. (4) Conclusions: Adapting adult speech has proven successful in child-speech synthesis. It appears that the resulting quality can be improved by training a large amount of pre-selected speech data, aided by a neural-network classifier, to better match the children’s speech. We encourage future research surrounding individualised synthetic speech for children with CCN, with special attention to children who make use of low-resource languages.

show abstract

Section: Speech-synthesis Systemsmentioning

confidence: 99%

Section: Speech-synthesis Systemsmentioning

confidence: 99%

Section: Speech-synthesis Systemsmentioning

confidence: 99%

Section: Child-speech Datamentioning

confidence: 99%

See 2 more Smart Citations

A Situational Analysis of Current Speech-Synthesis Systems for Child Voices: A Scoping Review of Qualitative and Quantitative Evidence

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Maximum likelihood linear regression (MLLR) adaptation approaches can improve results [8] but not sufficiently to approach corresponding adult models. Using VTLN as a prior for the MLLR family of adaptation transformations [12] has proven to be effective for adapting child speech in HMM speech synthesis.…”

Section: Introductionmentioning

confidence: 99%

Improving Children’s Speech Recognition Through Out-of-Domain Data Augmentation

Fainberg

Bell²,

Lincoln³

et al. 2016

Interspeech 2016

View full text Add to dashboard Cite

Children's speech poses challenges to speech recognition due to strong age-dependent anatomical variations and a lack of large, publicly-available corpora. In this paper we explore data augmentation for children's speech recognition using stochastic feature mapping (SFM) to transform out-of-domain adult data for both GMM-based and DNN-based acoustic models. We performed experiments on the English PF-STAR corpus, augmenting using WSJCAM0 and ABI. Our experimental results indicate that a DNN acoustic model for childrens speech can make use of adult data, and that out-of-domain SFM is more accurate than in-domain SFM.

show abstract

Morphological normalization of vowel images for articulatory speech recognition

Wei

Zhang

et al. 2016

Journal of Visual Communication and Image Representation

View full text Add to dashboard Cite

Combining Vocal Tract Length Normalization With Hierarchical Linear Transformations

Cited by 5 publications

References 22 publications

A Situational Analysis of Current Speech-Synthesis Systems for Child Voices: A Scoping Review of Qualitative and Quantitative Evidence

A Situational Analysis of Current Speech-Synthesis Systems for Child Voices: A Scoping Review of Qualitative and Quantitative Evidence

Improving Children’s Speech Recognition Through Out-of-Domain Data Augmentation

Morphological normalization of vowel images for articulatory speech recognition

Contact Info

Product

Resources

About