Adding filled pauses and disfluent events into language models for speech recognition

Staš, Ján; Hládek, Daniel; Juhar, Jozef

doi:10.1109/coginfocom.2016.7804538

Cited by 5 publications

(5 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the further research, we want to focus on eliminating common recognition errors by introducing unsupervised language model adaptation to the current topic and specific speaker speaking style and statistical modeling of most frequent hesitation fillers in spontaneous speech for improving system performance and automatic transcription and annotation of large acoustic corpora of the spoken Slovak language [28]. Also, we are planning to append the fully-annotated data from that corpus to the current training data in order to retrain the present acoustic and language models.…”

Section: Resultsmentioning

confidence: 99%

TEDxSK and JumpSK: A New Slovak Speech Recognition Dedicated Corpus

Staš

Hládek

Viszlay

et al. 2017

Journal of Linguistics/Jazykovedný Casopis

View full text Add to dashboard Cite

This paper describes a new Slovak speech recognition dedicated corpus built from TEDx talks and Jump Slovakia lectures. The proposed speech database consists of 220 talks and lectures in total duration of about 58 hours. Annotated speech database was generated automatically in an unsupervised manner by using acoustic speech segmentation based on principal component analysis and automatic speech transcription using two complementary speech recognition systems. The evaluation data consisting of 50 manually annotated talks and lectures in total duration of about 12 hours, has been created for evaluation of the quality of Slovak speech recognition. By unsupervised automatic annotation of TEDx talks and Jump Slovakia lectures we have obtained 21.26% of new speech segments with approximately 9.44% word error rate, suitable for retraining or adaptation of acoustic models trained beforehand.

show abstract

Section: Resultsmentioning

confidence: 99%

TEDxSK and JumpSK: A New Slovak Speech Recognition Dedicated Corpus

Staš

Hládek

Viszlay

et al. 2017

Journal of Linguistics/Jazykovedný Casopis

View full text Add to dashboard Cite

show abstract

“…In the same vein, our work focuses on modeling false starts in text. Most similar to our work, [4] introduces disfluencies to clean text and uses the processed text for the language model. They, however, use an existing set of disfluencies to be introduced in the text.…”

Section: Related Workmentioning

confidence: 95%

“…In the context of language modeling, quite a few researchers [3,4,17,18,19,20] have studied disfluencies like false starts, filled pauses and repetition in textual data. Some of these work [3,4] have noted that modeling disfluencies can be beneficial for language modeling. In the same vein, our work focuses on modeling false starts in text.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Data Augmentation Using Prosody and False Starts to Recognize Non-Native Children’s Speech

Kathania¹,

Singh²,

Grósz³

et al. 2020

Interspeech 2020

View full text Add to dashboard Cite

This paper describes AaltoASR's speech recognition system for the INTERSPEECH 2020 shared task on Automatic Speech Recognition (ASR) for non-native children's speech. The task is to recognize non-native speech from children of various age groups given a limited amount of speech. Moreover, the speech being spontaneous has false starts transcribed as partial words, which in the test transcriptions leads to unseen partial words. To cope with these two challenges, we investigate a data augmentation-based approach. Firstly, we apply the prosodybased data augmentation to supplement the audio data. Secondly, we simulate false starts by introducing partial-word noise in the language modeling corpora creating new words. Acoustic models trained on prosody-based augmented data outperform the models using the baseline recipe or the SpecAugment-based augmentation. The partial-word noise also helps to improve the baseline language model. Our ASR system, a combination of these schemes, is placed third in the evaluation period and achieves the word error rate of 18.71%. Post-evaluation period, we observe that increasing the amounts of prosody-based augmented data leads to better performance. Furthermore, removing low-confidence-score words from hypotheses can lead to further gains. These two improvements lower the ASR error rate to 17.99%.

show abstract

“…Generally speaking, human-computer speech communication has been one of the most popular topics in CogInfoCom research area. In paper [31], the authors discuss including filled pauses and disfluent events into the training data for statistical language modeling, in order to improve speech recognition accuracy and robustness in the case of spontaneous speech. In [30], speech analysis has been conducted to verify the speaker authorization and measure the stress level within the air-ground voice communication, to improve voice communication in air traffic management security.…”

Section: Related Workmentioning

confidence: 99%

Whispered Speech Recognition using Hidden Markov Models and Support Vector Machines

2018

APH

View full text Add to dashboard Cite

Whisper is a specific mode of speech characterized by turbulent airflow at the glottis level. Despite an increased effort in speech perception, the intelligibility of whisper in human communication is very high. An enormous acoustic mismatch between normally phonated (neutral) and whispered speech is the main reason why modern Automatic Speech Recognition (ASR) systems have significant drop of performances when applied to whisper. In this paper, we present an analysis in recognition of whisper using 2 machinelearning techniques: Hidden Markov Models (HMM) and Support Vector Machines (SVM). The experiments are conducted in both Speaker Dependent (SD) and Speaker Independent (SI) fashion for Whi-Spe speech database. The best neutral-trained whisper recognition accuracy in SD fashion (83.36%) is obtained in SVM framework. At the same time, HMMbased recognition gave the highest recognition accuracy in SI fashion (87.42%). The results in recognition of neutral speech are given as well.

show abstract

Adding filled pauses and disfluent events into language models for speech recognition

Cited by 5 publications

References 8 publications

TEDxSK and JumpSK: A New Slovak Speech Recognition Dedicated Corpus

TEDxSK and JumpSK: A New Slovak Speech Recognition Dedicated Corpus

Data Augmentation Using Prosody and False Starts to Recognize Non-Native Children’s Speech

Whispered Speech Recognition using Hidden Markov Models and Support Vector Machines

Contact Info

Product

Resources

About