A Comparison of Different Approaches to Automatic Speech Segmentation

Demuynck, Kris; Laureys, Tom

doi:10.1007/3-540-46154-x_38

Cited by 27 publications

(21 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This finding is of utmost importance as it allows achieving fair alignment rates on corpora for which a standard alignment performs very poorly. An example of such improvement can be observed on the 30 and 88.9 % with the use of VAD. Consistency is found across all small corpora, with a minimum relative improvement of 32.2 % and 15.8 % for 40 ms errors, with corpora of 30 and 60 seconds respectively.…”

Section: F Advantage Of Tanda2 In Small Corporamentioning

confidence: 86%

“…a regression tree [12], Gaussian Mixture Models (GMMs) or HMMs [28] or support vector machines (SVM) [29]) which requires a manuallyannotated corpus. Heuristic rules can also be used to modify the boundaries as a post-process, making use of the average deviation for each pair of phonemes [30]. In a similar way, it is proposed in [31] to train, on a manually-aligned corpus, boundary models depending on the right and left context of each phoneme to improve the alignment.…”

Section: Existing Techniquesmentioning

confidence: 99%

“…It is widely acknowledged that part of the errors made by automatic phonetic alignment techniques comes from the fact that humans make use of additional cues, which are not represented in the MFCC coefficients, to decide the precise location of the boundaries [30]. This may especially be the case for vowel-vowel transitions or transitions to and from semi-vowels which are very hard to model.…”

Section: B Augmentation Of the Feature Vector (Addfeat)mentioning

confidence: 99%

See 2 more Smart Citations

HMM-Based Speech Segmentation: Improvements of Fully Automatic Approaches

Brognaux

Drugman

2016

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Speech segmentation refers to the problem of determining the phoneme boundaries from an acoustic recording of an utterance together with its orthographic transcription. This paper focuses on a particular case of Hidden Markov Model (HMM) based forced alignment in which the models are directly trained on the corpus to align. The obvious advantage of this technique is that it is applicable to any language or speaking style and does not require manually-aligned data. Through a systematic stepby-step study, the role played by various training parameters (e.g. models configuration, number of training iterations) on the alignment accuracy is assessed, with corpora varying in speaking style and language. Based on a detailed analysis of the errors commonly made by this technique, we also investigate the use of additional fully automatic strategies to improve the alignment. Beside the use of supplementary acoustic features, we explore two novel approaches: an initialization of the silence models based on a Voice Activity Detection (VAD) algorithm and the consideration of the forced alignment of the timereversed sound. The evaluation is carried out on 12 corpora of different sizes, languages (some being under-resourced) and speaking styles. It aims at providing a comprehensive study of the alignment accuracy achieved by the different versions of the speech segmentation algorithm depending on corpus-related specificities. While the baseline method is shown to reach good alignment rates with corpora as small as 2 minutes, we also emphasize the benefit of using a few seconds of bootstrapping data. Regarding improvement methods, our results show that the insertion of additional features outperforms both other strategies. The performance of VAD, however, is shown to be notably striking on very small corpora, correcting more than 60 % of the errors superior to 40 ms. Finally, the combination of the three improvement methods is also pointed out as providing the highest alignment rates, with very low variability across the corpora, regardless of their size. This combined technique is shown to outperform available speaker-independent models, improving the alignment rate by 8 to 10 % absolute.

show abstract

Section: F Advantage Of Tanda2 In Small Corporamentioning

confidence: 86%

Section: Existing Techniquesmentioning

confidence: 99%

Section: B Augmentation Of the Feature Vector (Addfeat)mentioning

confidence: 99%

See 1 more Smart Citation

HMM-Based Speech Segmentation: Improvements of Fully Automatic Approaches

Brognaux

Drugman

2016

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…The forward-backward segmentation [6], proposed as an alternative to the Viterbi algorithm led to good results in condition that the Viterbi segmentation was started with manual segmentation.…”

mentioning

confidence: 93%

Optimized segmentation method of the vocal signal with predictive neural network

Zbancioc

Feraru

2014

Proceedings of the 2014 6th International Conference on Electronics, Computers and Artificial Intelligence (ECAI)

View full text Add to dashboard Cite

This paper presents an efficient segmentation method which uses simple neural network architecture. The goal was to implement an automatic annotation instrument of the vocal signal, capable to make the separation between vowel and consonant signal, respectively pauses between utterances. This instrument is used in the features extraction from the vowels areas by an emotion recognition application for the Romanian language.

show abstract

“…by means of time stamps [1]. Phonemic segmentation is commonly used in preprocessing and initial training steps of automatic speech/speaker recognition, speech enhancement, computer-aided speech transcription systems, and the development of corpus-based speech synthesis systems.…”

mentioning

confidence: 99%

Robust Detection of Phone Boundaries Using Model Selection Criteria With Few Observations

Almpanidis

Kotti

Kotropoulos

2009

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Abstract-Automatic phone segmentation techniques based on model selection criteria are studied. We investigate the phone boundary detection efficiency of entropy-and Bayesian-based model selection criteria in continuous speech based on the DISTBIC hybrid segmentation algorithm. DISTBIC is a text-independent bottom-up approach that identifies sequential model changes by combining metric distances with statistical hypothesis testing. Using robust statistics and small sample corrections in the baseline DISTBIC algorithm, phone boundary detection accuracy is significantly improved, while false alarms are reduced. We also demonstrate further improvement in phonemic segmentation by taking into account how the model parameters are related in the probability density functions of the underlying hypotheses as well as in the model selection via the information complexity criterion and by employing M-estimators of the model parameters. The proposed DISTBIC variants are tested on the NTIMIT database and the achieved 1 measure is 74.7% using a 20-ms tolerance in phonemic segmentation.

show abstract

A Comparison of Different Approaches to Automatic Speech Segmentation

Cited by 27 publications

References 2 publications

HMM-Based Speech Segmentation: Improvements of Fully Automatic Approaches

HMM-Based Speech Segmentation: Improvements of Fully Automatic Approaches

Optimized segmentation method of the vocal signal with predictive neural network

Robust Detection of Phone Boundaries Using Model Selection Criteria With Few Observations

Contact Info

Product

Resources

About