Speech Recognition Research Articles

Application of Deep Learning Technology in Speech Recognition and Language Teaching

In the process of learning a second language, modern Chinese learners are faced with many challenges due to the limitations of the learning environment and teaching conditions. The teachers are usually non-native speakers, and their language level is uneven, resulting in unsatisfactory teaching results. At the same time, the size of the group learning a second language continues to expand, and the requirement for teaching level is getting higher and higher. The topic of this paper is the application of deep learning to speech recognition and language teaching. The results and effects of deep learning in speech recognition and language teaching are summarized by combining the currently available information. It can be found that the current deep learning technology is more and more advanced, the accuracy rate of speech recognition technology has been greatly improved compared with the past, and the teaching effect has also made no small achievements. This technology can not only help learners better master English pronunciation, but also provide more efficient and effective learning experience.

Integrating international Chinese visualization teaching and vocational skills training: leveraging attention-connectionist temporal classification models.

The teaching of Chinese as a second language has become increasingly crucial for promoting cross-cultural exchange and mutual learning worldwide. However, traditional approaches to international Chinese language teaching have limitations that hinder their effectiveness, such as outdated teaching materials, lack of qualified instructors, and limited access to learning facilities. To overcome these challenges, it is imperative to develop intelligent and visually engaging methods for teaching international Chinese language learners. In this article, we propose leveraging speech recognition technology within artificial intelligence to create an oral assistance platform that provides visualized pinyin-formatted feedback to learners. Additionally, this system can identify accent errors and provide vocational skills training to improve learners' communication abilities. To achieve this, we propose the Attention-Connectionist Temporal Classification (CTC) model, which utilizes a specific temporal convolutional neural network to capture the location information necessary for accurate speech recognition. Our experimental results demonstrate that this model outperforms similar approaches, with significant reductions in error rates for both validation and test sets, compared with the original Attention model, Claim, Evidence, Reasoning (CER) is reduced by 0.67%. Overall, our proposed approach has significant potential for enhancing the efficiency and effectiveness of vocational skills training for international Chinese language learners.

Exploration of the Application of Artificial Intelligence in High School English Speaking Teaching

With the remarkable strides in artificial intelligence (AI) technology, a range of sophisticated techniques and products have sprung forth, including speech recognition and semantic analysis. These advancements have propelled the application of AI to new heights, particularly in the realm of education. Notably, AI applications such as speech recognition, human-computer interaction, and automated evaluation have revolutionized traditional English speaking teaching, introducing unprecedented changes to the way students learn and interact. The paper aims to investigate how AI might enhance high school pupils' English-speaking proficiency, which employs a literature review methodology to examine and illustrate the smooth integration of AI technology with high school English speaking instruction. This approach enhances speaking learning materials and fosters creativity in teaching strategies and assessment techniques. In addition, it creates room for the evolution of the oral teaching style and broadens the learning environment. This intelligent transformation of the oral teaching mode is expected to effectively enhance the effectiveness of high school oral English teaching and ultimately improve students' language proficiency.

Comparison and analysis of new curriculum criteria for end-to-end ASR

Traditionally, teaching a human and a Machine Learning (ML) model is quite different, but organized and structured learning has the ability to enable faster and better understanding of the underlying concepts. For example, when humans learn to speak, they first learn how to utter basic phones and then slowly move towards more complex structures such as words and sentences. Motivated by this observation, researchers have started to adapt this approach for training ML models. Since the main concept, the gradual increase in difficulty, resembles the notion of the curriculum in education, the methodology became known as Curriculum Learning (CL). In this work, we design and test new CL approaches to train Automatic Speech Recognition systems, specifically focusing on the so-called end-to-end models. These models consist of a single, large-scale neural network that performs the recognition task, in contrast to the traditional way of having several specialized components focusing on different subtasks (e.g., acoustic and language modeling). We demonstrate that end-to-end models can achieve better performances if they are provided with an organized training set consisting of examples that exhibit an increasing level of difficulty. To impose structure on the training set and to define the notion of an easy example, we explored multiple solutions that use either external, static scoring methods or incorporate feedback from the model itself. In addition, we examined the effect of pacing functions that control how much data is presented to the network during each training epoch. Our proposed curriculum learning strategies were tested on the task of speech recognition on two data sets, one containing spontaneous Finnish speech where volunteers were asked to speak about a given topic, and one containing planned English speech. Empirical results showed that a good curriculum strategy can yield performance improvements and speed-up convergence. After a given number of epochs, our best strategy achieved a 5.6% and 3.4% decrease in terms of test set word error rate for the Finnish and English data sets, respectively.

Development of 20-Talker Korean Sentence In Noise Test Lists

Purpose: Clinical speech recognition measures often present the sentences recorded with a single speaker and one rate of speech. Performances with single-talker recordings may not accurately represent the speech recognition abilities of the listeners in a multitalker communication situation. The present study aimed to construct and optimize the sentences recorded by 20 different talkers for sentence-in-noise recognition tests (20-talker Korean sentence-in-noise test, 20-talker K-SIN).Methods: Phases I and II were conducted in this study. In phase I (developmental phase), preliminary 720 sentences composed of 3 to 6 words were selected and recorded by twenty different talkers (10 male and 10 female). The recorded sentences were superimposed to generate a speech-shaped noise similar to the long-term average speech spectrum of the sentences. In phase II (optimization and formation of equally intelligible sentence lists), the psychometric functions of 30 normal-hearing listeners were obtained from the sentence-in-noise recognition scores at three different signal-to-noise ratios (SNRs) (-2, -4, and -7 dB). Based on these scores, the SNR required for 50% sentence intelligibility (SNR-50) and the slope at that point were derived from the psychometric function curves.Results: Before level adjustment, the median SNR-50 and slope were -4.12 dB SNR and 24.33%/dB over 720 sentences. Level adjustments were applied to homogenize the intelligibility of the sentences, resulting in 508 sentences remained. Out of 508 sentences, 320 sentences were used to construct 1 practice list and 15 test lists, wherein 20 sentences in each list were spoken by each of the 20 different talkers.Conclusion: The 20-talker K-SIN sentences can be used for high-variability speech-in-noise recognition test to better reflect the multitalker communication abilities of listeners.

A Bibliometric Approach and Meta-Analysis of Effects of Automatic Speech Recognition on Second Language Learning

Although automatic speech recognition technologies have gained popularity in second language learning, there are still difficulties facing the use of automatic speech recognition in second language learning. Through a bibliometric review and a meta-analysis, this study examines the effects of automatic speech recognition on second language learning. The results of this investigation show that automatic speech recognition technologies can improve second language speaking skills and reduce speaking anxiety, but whether the listening skills can be improved relies on the degree of advancement of the automatic speech recognition technologies. Teachers and students positively evaluate the use of automatic speech recognition technologies in second language learning. Bibliometric analyses reveal the top ten authors, organizations, countries, cited sources, and cited authors, as well as the chronological research trend in the research on the automatic speech recognition used in second language education.

Enhancing customer experience through AI-driven language processing in service interactions

In today’s competitive service industries, delivering exceptional customer experiences is paramount. This paper explores how AI-driven language processing technologies enhance customer service interactions. Key technologies such as Natural Language Processing (NLP), chatbots, virtual assistants, speech recognition, and machine learning are crucial in transforming customer service. These technologies enable personalized, efficient, and consistent customer interactions, significantly improving overall customer satisfaction. However, implementing these technologies has challenges, including data quality, system integration, privacy concerns, user acceptance, and financial investment. Addressing these challenges is essential for successful AI adoption. Future trends indicate further advancements in AI capabilities, integration with IoT and blockchain technologies, and evolving ethical and regulatory considerations. The long-term impact of AI on service industries promises more efficient operations, cost savings, and superior customer experiences, giving companies a competitive advantage in the market. This paper provides insights into leveraging AI to enhance customer experience, outlining the benefits and challenges while speculating on future directions and implications.

Individual differences in visual pattern completion predict adaptation to degraded speech

Recognizing acoustically degraded speech relies on predictive processing whereby incomplete auditory cues are mapped to stored linguistic representations via pattern recognition processes. While listeners vary in their ability to recognize degraded speech, performance improves when a written transcription is presented, allowing completion of the partial sensory pattern to preexisting representations. Building on work characterizing predictive processing as pattern completion, we examined the relationship between domain-general pattern recognition and individual variation in degraded speech learning. Participants completed a visual pattern recognition task to measure individual-level tendency towards pattern completion. Participants were also trained to recognize noise-vocoded speech with written transcriptions and tested on speech recognition pre- and post-training using a retrieval-based transcription task. Listeners significantly improved in recognizing speech after training, and pattern completion on the visual task predicted improvement for novel items. The results implicate pattern completion as a domain-general learning mechanism that can facilitate speech adaptation in challenging contexts.

Imperceptible backdoor watermarks for speech recognition model copyright protection

As the demand for shared pre-trained deep neural network models continues to rise, safeguarding the intellectual property of models is also increasingly significant. While existing studies predominantly concentrate on protecting pre-trained image recognition models, limited research covers pre-trained speech recognition models. In this paper, we propose the black-box watermark method to authenticate the ownership of speech recognition models. This method can mitigate the risk of unauthorized AI services being created by attackers who gain access to the pre-trained model. Accordingly, we present three watermarking methods: Gaussian noise watermark, extreme frequency Gaussian noise watermark, and unrelated audio watermark. These generated watermarks, embedded into models through training or fine-tuning, exhibit remarkable fidelity and effectiveness, backed by rigorous experimental validation. Furthermore, our experiments reveal that the extreme frequency noise backdoor enhances the robustness of the watermark compared to the Gaussian noise and unrelated audio watermark.

Factors to Describe the Outcome Characteristics of a CI Recipient.

Background: In cochlear implant (CI) treatment, there is a large variability in outcome. The aim of our study was to identify the independent audiometric measures that are most directly relevant for describing this variability in outcome characteristics of CI recipients. An extended audiometric test battery was used with selected adult patients in order to characterize the full range of CI outcomes. Methods: CI users were recruited for this study on the basis of their postoperative results and divided into three groups: low (1st quartile), moderate (medium decentile), and high hearing performance (4th quartile). Speech recognition was measured in quiet by using (i) monosyllabic words (40-80 dB SPL), (ii) speech reception threshold (SRT) for numbers, and (iii) the German matrix test in noise. In order to reconstruct demanding everyday listening situations in the clinic, the temporal characteristics of the background noise and the spatial arrangements of the signal sources were varied for tests in noise. In addition, a survey was conducted using the Speech, Spatial, and Qualities (SSQ) questionnaire and the Listening Effort (LE) questionnaire. Results: Fifteen subjects per group were examined (total N = 45), who did not differ significantly in terms of age, time after CI surgery, or CI use behavior. The groups differed mainly in the results of speech audiometry. For speech recognition, significant differences were found between the three groups for the monosyllabic tests in quiet and for the sentences in stationary (S0°N0°) and fluctuating (S0°NCI) noise. Word comprehension and sentence comprehension in quiet were both strongly correlated with the SRT in noise. This observation was also confirmed by a factor analysis. No significant differences were found between the three groups for the SSQ questionnaire and the LE questionnaire results. The results of the factor analysis indicate that speech recognition in noise provides information highly comparable to information from speech intelligibility in quiet. Conclusions: The factor analysis highlighted three components describing the postoperative outcome of CI patients. These were (i) the audiometrically measured supra-threshold speech recognition and (ii) near-threshold audibility, as well as (iii) the subjective assessment of the relationship to real life as determined by the questionnaires. These parameters appear well suited to setting up a framework for a test battery to assess CI outcomes.

Cochlear Implantation in Single-Sided Deafness and Asymmetric Hearing Loss: 12 Months Follow-up Results of a European Multicenter Evaluation.

People with single-sided deafness (SSD) or asymmetric hearing loss (AHL) have particular difficulty understanding speech in noisy listening situations and in sound localization. The objective of this multicenter study is to evaluate the effect of a cochlear implant (CI) in adults with single-sided deafness (SSD) or asymmetric hearing loss (AHL), particularly regarding sound localization and speech intelligibility with additional interest in electric-acoustic pitch matching. A prospective longitudinal study at 7 European tertiary referral centers was conducted including 19 SSD and 16 AHL subjects undergoing cochlear implantation. Sound localization accuracy was investigated in terms of root mean square error and signed bias before and after implantation. Speech recognition in quiet and speech reception thresholds in noise for several spatial configurations were assessed preoperatively and at several post-activation time points. Pitch perception with CI was tracked using pitch matching. Data up to 12 months post activation were collected. In both SSD and AHL subjects, CI significantly improved sound localization for sound sources on the implant side, and thus overall sound localization. Speech recognition in quiet with the implant ear improved significantly. In noise, a significant head shadow effect was found for SSD subjects only. However, the evaluation of AHL subjects was limited by the small sample size. No uniform development of pitch perception with the implant ear was observed. The benefits shown in this study confirm and expand the existing body of evidence for the effectiveness of CI in SSD and AHL. Particularly, improved localization was shown to result from increased localization accuracy on the implant side.

What makes a medical translator?

Advancements in healthcare and the rise of patient-centredness have increased workloads for professional medical translators. In parallel, new and emerging technologies are impacting translators’ output and workflows, though investigations into the medical translation domain remain scarce. This article presents the results of an international survey on 165 professional medical translators to identify trends concerning their socio-demographic and professional profiles, the textual- and process-related challenges of translating medical texts aimed at patients as laypersons, and their use of technology, specifically Computer-Assisted Translation (CAT) tools and Automatic Speech Recognition (ASR). Results show that the medical translators surveyed have decades-long experience in translation and come primarily from a linguistic background. They find specialised terminology and delivery deadlines as most challenging when translating patient-facing medical texts. Finally, most medical translators in our survey use CAT tools when translating or post-editing, although a small minority also integrate ASR tools into their workflows.

Application of speech recognition algorithm based on interactive artificial intelligence system in English video teaching system

Social robots are intelligent programs that can autonomously run, automatically publish information, and interact on social media platforms. This article focuses on the speech recognition algorithm of interactive artificial intelligence systems, combined with the Hanning window function and the Vitit algorithm, and explores its application value in English video teaching systems through detailed experiments. Adopting speech feature extraction technology provides strong support for the subsequent recognition process. The application of Hanning window function has a positive impact on speech signal preprocessing. The most indispensable algorithm in speech recognition models is the Vitit algorithm, which successfully solves the problem of time variation and continuity of speech signals. This study successfully applied speech recognition algorithms based on interactive artificial intelligence systems to English video teaching systems. By combining speech feature extraction, Hanning window function, and Vitit algorithm, a system with strong recognition ability and interactivity was constructed. The English video teaching system aims to improve students’ learning process. By combining interactive artificial intelligence speech recognition technology with English video teaching, it can improve students’ experience in learning English and also meet their personalized learning habits.

An automatic speech recognition system in Odia language using attention mechanism and data augmentation

An automatic speech recognition system in Odia language using attention mechanism and data augmentation

Adaptive unified defense framework for tackling adversarial audio attacks

Adversarial attacks aimed at subverting recognition systems have laid bare significant security vulnerabilities inherent in deep neural networks. In the automatic speech recognition (ASR) domain, prevailing defense mechanisms have primarily centered on pre-processing procedures to mitigate adversarial threats stemming from such attacks. However, despite their initial success, these methods have shown surprising vulnerabilities when confronted with robust and adaptive adversarial attacks. This paper proposes an adaptive unified defense framework tailored to address the challenges posed by robust audio adversarial examples. The framework comprises two pivotal components: (1) a unified pre-processing mechanism is designed to disrupt the continuity and transferability of adversarial attacks. Its objective is to thwart the consistent operation of adversarial examples across different systems or conditions, thereby enhancing the robustness of the defense. (2) an adaptive ASR transcription method is proposed to further bolster our defense strategy. Empirical experiments conducted using two benchmark audio datasets within a state-of-the-art ASR system affirm the effectiveness of our adaptive defense framework. It achieves an impressive 100% accuracy rate against representative audio attacks and consistently outperforms other state-of-the-art defense techniques, achieving an accuracy rate of 98.5% even when faced with various challenging adaptive adversarial attacks.

Effect of Human Interaction Modalities on the Usability and Performance of a Robotic Cyber-Physical System for Construction Progress Monitoring

The construction industry has been slow to adopt advanced technologies like Cyber-Physical Systems (CPS), underscoring the necessity for a human-centered approach to their development. This study aims to understand the significance and impact of human-CPS interaction modalities on human perceptions of CPS, specifically in the use case of construction progress monitoring. A mixed-method approach comprising quantitative observations and qualitative interviews was employed to evaluate the system’s usability and performance with different human-CPS interaction modalities. The findings reveal variable levels of usability and performance across different human-CPS interaction modalities. Users expressed a preference for the three-dimensional immersive environment compared to a two-dimensional setting due to a better sense of presence. However, they experienced heightened cognitive load when utilizing novel interaction methods such as hand gestures and voice commands. Interestingly, despite this increased cognitive demand, users exhibited comparable performance in the assigned task, irrespective of their familiarity with the interaction modality. The results underscore the importance of human-centered design in CPS development. It contributes empirical evidence crucial for understanding and improving human-CPS interactions, thereby advancing the implementation of CPS in construction progress monitoring.

Building Automatic Speech Recognition Systems for Moroccan Dialect: A Phoneme-Based Approach

Building Automatic Speech Recognition Systems for Moroccan Dialect: A Phoneme-Based Approach

Speech Recognition Technology for Nursing Charting: A Literature Review.

This literature review explores the impact of Speech Recognition Technology (SRT) on nursing documentation within electronic health records (EHR). A search across PubMed, CINAHL, and Google Scholar identified 156 studies, with seven meeting the inclusion criteria. These studies investigated the impact of SRT on documentation time, accuracy, and user satisfaction. Findings suggest SRT, particularly when integrated with artificial intelligence can speed up documentation by up to 15%. However, challenges remain in its implementation in real-world clinical settings and existing EHR workflows. Future studies should focus on developing SRT systems that process conversational nursing assessments and integrate into current EHRs.

Grammatical gender in spoken word recognition in school-age Spanish monolingual and Spanish-English bilingual children.

This study examined grammatical gender processing in school-aged children with varying levels of cumulative English exposure. Children participated in a visual world paradigm with a four-picture display where they heard a gendered article followed by a target noun and were in the context where all images were the same gender (same gender), where all of the distractor images were the opposite gender than the target noun (different gender), and where all of the distractor images were the opposite gender, but there was a mismatch in the gendered article and target noun pair. We investigated 51 children (aged 5;0-10;0) who were exposed to Spanish since infancy but varied in their amount of cumulative English exposure. In addition to the visual word paradigm, all children completed an article-noun naming task, a grammaticality judgment task, and standardized vocabulary tests. Parents reported on their child's cumulative English language exposure and current English language use. To investigate the time course of lexical facilitation effects, looks to the target were analyzed with a cluster-based permutation test. The results revealed that all children used gender in a facilitatory way (during the noun region), and comprehension was significantly inhibited when the article-noun pairing was ungrammatical rather than grammatical. Compared to children with less cumulative English exposure, children with more cumulative English exposure looked at the target noun significantly less often overall, and compared to younger children, older children looked at the target noun significantly more often overall. Additionally, children with lower cumulative English exposure looked at target nouns more in the different-gender condition than the same-gender condition for masculine items more than feminine items.

A Pilot Implementation and Evaluation of Speech Recognition Technology for Nurses and Health Disciplines.

This pilot study addresses the pervasive issue of burnout among nurses and health disciplines, often exacerbated by the use of electronic health record (EHR) systems. Recognizing the potential of dictation to alleviate documentation burden, the study focuses on the adoption of speech recognition technology (SRT) in a large Canadian urban mental health and addiction teaching hospital. Clinicians who participated in the pilot provided feedback on their experiences via a survey, and analytics data were examined to measure usage and adoption patterns. Preliminary feedback reveals a subset of participants rapidly embracing the technology, reporting decreased documentation times and increased efficiency. However, some clinicians experienced challenges related to initial setup time and the effort of adjusting to a novel documentation approach.