Estimating the quality of phonetic transcriptions and segmentations of speech signals

Wesenick, Maria-Barbara; Kipp, Andreas

doi:10.1109/icslp.1996.607054

Cited by 30 publications

(22 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…These differences are also similar to those typically obtained between human transcribers. For instance, Wesenick and Kipp (1996) reported that 96% of the segment boundaries determined by three human transcribers for 64 read sentences of German differed by less than 20 ms. Similarly, Raymond and colleagues (Raymond, Pitt, Johnson, Hume, Makashay, Dautricourt, & Hilts, 2002) observed an average alignment difference of 16.4 ms between human transcribers of the Buckeye corpus.…”

Section: Methodsmentioning

confidence: 99%

“…We automatically created broad phonetic transcriptions for all speech in the corpus, using the transcriber and procedure described in Schuppler et al (2011). This automatic transcriber is based on the Hidden Markov Toolkit HTK (Young et al, 2002), and received as its input the acoustic signals and the corresponding orthographic transcriptions. The recognizer selected for each word in the orthographic transcriptions the pronunciation variant that best matched the acoustic signal, choosing from a lexicon containing for every word both the full pronunciation and several reduced pronunciation variants.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

The effect of speech situation on the occurrence of reduced word pronunciation variants

Ernestus

Hanique

Verboom

2015

Journal of Phonetics

View full text Add to dashboard Cite

This article presents two studies investigating how the situation in which speech is uttered affects the frequency with which words are reduced. Study 1 is based on the Spoken Dutch Corpus, which consists of 15 components, nearly all representing a different speech situation. This study shows that the components differ in how often ten semantically weak words are highly reduced. The differences are especially large between the components with scripted and unscripted speech. Within the component group of unscripted speech, the formality of the situation shows an effect. Study 2 investigated segment reduction in a shadowing experiment in which participants repeated Dutch carefully and casually articulated sentences. Prefixal schwa and suffixal /t/ were absent in participants' responses to both sentences types as often as in formal interviews. If a segment was absent, this appeared to be mostly due to extreme coarticulation, unlike in speech produced in less formal situations. Speakers thus adapted more to the formal situation of the experiment than to the stimuli to be shadowed. We conclude that speech situation affects the occurrence of reduced word pronunciation variants, which should be accounted for by psycholinguistic models of speech production and comprehension.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

The effect of speech situation on the occurrence of reduced word pronunciation variants

Ernestus

Hanique

Verboom

2015

Journal of Phonetics

View full text Add to dashboard Cite

show abstract

“…These average differences are similar to those obtained in other studies. For example, Wesenick and Kipp (1996) reported that 96% of the segment boundaries determined by three human transcribers for 64 read sentences of German differ by less than 20 ms. Similarly, Raymond et al (2002) presented an average alignment difference of 16.4 ms between human transcribers of the Buckeye corpus.…”

Section: Measurementsmentioning

confidence: 99%

Informal speech processes can be categorical in nature, even if they affect many different words

Hanique

Ernestus

Schuppler

2013

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

This paper investigates the nature of reduction phenomena in informal speech. It addresses the question whether reduction processes that affect many word types, but only if they occur in connected informal speech, may be categorical in nature. The focus is on reduction of schwa in the prefixes and on word-final /t/ in Dutch past participles. More than 2000 tokens of past participles from the Ernestus Corpus of Spontaneous Dutch and the Spoken Dutch Corpus (both from the interview and read speech component) were transcribed automatically. The results demonstrate that the presence and duration of /t/ are affected by approximately the same phonetic variables, indicating that the absence of /t/ is the extreme result of shortening, and thus results from a gradient reduction process. Also for schwa, the data show that mainly phonetic variables influence its reduction but its presence is affected by different and more variables than its duration, which suggests that the absence of schwa may result from gradient as well as categorical processes. These conclusions are supported by the distributions of the segments' durations. These findings provide evidence that reduction phenomena which affect many words in informal conversations may also result from categorical reduction processes.

show abstract

“…It should be noted that several studies have pointed out the high degree of inter-rater disagreement, with sometimes large discrepancies between human-made alignments [37]. Generally, results are provided within a certain tolerance threshold on the timing error.…”

Section: B Evaluation Metricsmentioning

confidence: 97%

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

Estimating the quality of phonetic transcriptions and segmentations of speech signals

Cited by 30 publications

References 3 publications

The effect of speech situation on the occurrence of reduced word pronunciation variants

The effect of speech situation on the occurrence of reduced word pronunciation variants

Informal speech processes can be categorical in nature, even if they affect many different words

HMM-Based Speech Segmentation: Improvements of Fully Automatic Approaches

Contact Info

Product

Resources

About