Impact of pronunciation variation in speech recognition

Brunet, R. Golda; Murthy, Hema A.

doi:10.1109/spcom.2012.6290037

Cited by 5 publications

(3 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ANOVA (analysis of variance) test is used to study the effect of different social factors on the above-mentioned rhythm metrics [26]. In addition, Post hoc tests are performed to understand the sources of those differences.…”

Section: Discussionmentioning

confidence: 99%

On the relevance of speech rhythm metrics to characterize social factors

Ayed

Hamdani-Droua²,

Alotaibi

et al. 2013

2013 7th IEEE GCC Conference and Exhibition (GCC)

View full text Add to dashboard Cite

Previous work on rhythm in speech signal has attempted to quantify rhythm metrics and classify languages accordingly into different rhythm classes. These attempts have met with more or less success; new studies conducted on some varieties and regional dialects provide results that are not consistent with their corresponding rhythm class languages. In this work, we try to explain previous results by investigating to what extent rhythm metrics are conditioned by social factors such as differences in age, gender, region, and education level. Experiments conducted on two sentences of the TIMIT database spoken by 630 speakers show that rhythm metrics are highly sensitive to the first three above-mentioned factors.

show abstract

Section: Discussionmentioning

confidence: 99%

On the relevance of speech rhythm metrics to characterize social factors

Ayed

Hamdani-Droua²,

Alotaibi

et al. 2013

2013 7th IEEE GCC Conference and Exhibition (GCC)

View full text Add to dashboard Cite

show abstract

“…Adding pronunciation variants to a dictionary is controversial, as it can be considered advantageous or detrimental to the system. Brunet and Murthy [4] investigated the effect of pronunciation variation on recognition accuracy. In their work, a data-driven approach was employed to observe pronunciation variation at syllable level.…”

Section: Introductionmentioning

confidence: 99%

Bilateral G2P accuracy: Measuring the effect of variants

Giwa

Davel

2017

2017 Pattern Recognition Association of South Africa and Robotics and Mechatronics (PRASA-RobMech)

View full text Add to dashboard Cite

Incorporating pronunciation variants in a dictionary is controversial, as this can be either advantageous or detrimental for a speech recognition system. Grapheme-tophoneme (G2P) accuracy can help guide this decision, but calculating the G2P accuracy of variant-based dictionaries is not fully straightforward. We propose a variant matching technique to measure G2P accuracy in a principled way, when both the reference and hypothesised dictionaries may include variants. We use the new measure to evaluate G2P accuracy and speech recognition performance of systems developed with an existing set of dictionaries, and observe a better correlation between G2P accuracy and speech recognition performance, than when utilising alternative metrics.

show abstract

“…While incorporating necessary variants can improve both automatic speech recognition (ASR) and spoken term detection (STD) performance [1,2], introducing superfluous variants can lead to increased confusability and a decrease in performance [3]. The use of pronunciation variants in state-of-the-art speech-recognition systems has received significant attention in recent years, with a number of studies investigating the effect of variant prediction on either ASR or STD accuracy [4,5,1,3,6]. We review the most relevant of these below (in Section 2).…”

Section: Introductionmentioning

confidence: 99%

G2p variant prediction techniques for ASR and STD

2013

View full text Add to dashboard Cite

Introducing pronunciation variants into a lexicon is a balancing act: incorporating necessary variants can improve automatic speech recognition (ASR) and spoken term detection (STD) performance by capturing some of the variability that occurs naturally; introducing superfluous variants can lead to increased confusability and a decrease in performance. We experiment with two very different grapheme-to-phoneme variant prediction techniques and analyze the variants generated, as well as their effect when used within fairly standard ASR and STD systems with unweighted lexicons. Specifically, we compare the variants generated by joint sequence models, which use probabilistic information to generate as many or as few variants as required, with a more discrete approach: the use of pseudophonemes within the default-and-refine algorithm. We evaluate results using three of the 2013 Babel evaluation languages with quite different variant characteristics -Tagalog, Pashto and Turkish -and find that there are clear trends in how the number and type of variants influence performance, and that the implications for lexicon creation for ASR and STD are different.

show abstract

Impact of pronunciation variation in speech recognition

Cited by 5 publications

References 4 publications

On the relevance of speech rhythm metrics to characterize social factors

On the relevance of speech rhythm metrics to characterize social factors

Bilateral G2P accuracy: Measuring the effect of variants

G2p variant prediction techniques for ASR and STD

Contact Info

Product

Resources

About