Automatic Speech Recognition Output Research Articles

Monitoring student behavior in autonomous automatic speech recognition-based pronunciation practice

A growing number of studies are investigating the role of Automatic Speech Recognition (ASR) technology in second language (L2) pronunciation learning, highlighting its potential to promote autonomous learning and enhance the development of L2 pronunciation. However, how learners interact with ASR systems remains largely unknown. The current study examined the behavior of 46 English-as-a-foreign language learners over six sessions of autonomous ASR-based pronunciation practice. It also explored potential relationship between learners' behavior and individual differences in motivation and pronunciation attitude. Analyses of the learners' screen recordings shed light on their interaction with the ASR output, including the proportion of words successfully transcribed on the first attempt and unsuccessful cases requiring subsequent attempts—whether repaired, ignored, or remaining unrepaired despite several attempts. Results from motivation and attitude questionnaires revealed that learners with a higher pronunciation attitude at the study's onset were more likely to reattempt a word not successfully transcribed by the ASR system. Motivation, however, was not found to be linked to learner behavior. The findings contribute to the expanding research on ASR technology in L2 pronunciation learning by emphasizing the influence of learner behaviors and attitudes, and providing important insights for L2 practitioners and researchers.

Is the Same Performance Really the Same?: Understanding How Listeners Perceive ASR Results Differently According to the Speaker's Accent

Research suggests that automatic speech recognition (ASR) systems, which automatically convert speech to text, show different performances according to various input classes (e.g., accent, age), requiring attention to building fairer AI systems that would perform similarly across various input classes. However, would an AI system with the same performance regardless of input classes really be perceived as fair enough? To this end, we investigate how listeners perceive the ASR system of the same result differently according to whether the speaker is a native speaker (NS) or a non-native speaker (NNS), which may lead to unfair situations. We conducted a study (n = 420), where participants were given one of the ten speech recordings with various accents of the same script along with the same captions. We found that even with the same ASR output, listeners perceive the ASR results differently. They found captions to be more useful for NNS's speech and blamed NNS more for the errors than NS. Based on the findings, we present design implications suggesting that we should take a step further than just achieving the same performance across various input classes to build a fair ASR system.

Enhancing Armenian Automatic Speech Recognition Performance: A Comprehensive Strategy for Speed, Accuracy, and Linguistic Refinement

This research introduces a comprehensive strategy to enhance the performance of an existing automatic speech recognition (ASR) model, which has been previously documented in published articles. The study sets out to achieve several objectives. Firstly, it concentrates on updating the ASR model by retraining it with new datasets. This involves integrating samples from the latest Common Voice corpus release and data collected independently via the armspeech.com web application. Another key focus lies in optimizing the ASR model for near-real-time processing, intending to improve its speed and efficiency. The proposed adjustments to the model’s architecture aim to balance accuracy and processing speed, which is essential for applications requiring prompt speech recognition. Furthermore, the research explores the integration of Transformer models into the post-processing pipeline to introduce punctuation and capitalization into the ASR output. This step not only enhances the linguistic quality of transcriptions but also improves their readability and usability. In tandem with these advancements, the research presents a systematic approach to gathering, annotating, and storing datasets specifically tailored for punctuation and capitalization tasks. The methodology outlines the acquisition and organization of a dataset conducive to training Transformer models for these linguistic tasks. This comprehensive approach, which encompasses dataset enrichment, architectural modifications, and post-processing enhancements, aims to elevate the ASR model’s accuracy, speed, and linguistic refinement, with a particular focus on addressing the intricacies of the Armenian language. The research contributes valuable insights into the optimization of ASR systems, tackling both language-specific challenges and broader issues related to linguistic post-processing.

Alignment in ASR and L1 Listeners’ Recognition of L2 Learner Speech: French EFL Learners & Dictation.Io

This study is an extension of Inceoglu et al.’s (2023) study on Google Voice Typing as a pronunciation learning tool. We used the Automatic Speech Recognition (ASR) tool on the dictation.io website (Agarwal, 2022), and our participants were L2 English learners of a different L1, but similar proficiency level. Twelve L1 English listeners assessed the L2 English from four L1 French speakers in terms of intelligibility and comprehensibility, measured by word transcription and Likert scale ratings respectively. Their scores were compared to ASR output. The goal was to determine how accurate the tool is, and to what extent its accuracy correlates with human listeners. The results were generally consistent with those of Inceoglu et al. (2023), with few exceptions which we discuss in the current study.

Improving Readability for Automatic Speech Recognition Transcription

Modern Automatic Speech Recognition (ASR) systems can achieve high performance in terms of recognition accuracy. However, a perfectly accurate transcript still can be challenging to read due to grammatical errors, disfluency, and other noises common in spoken communication. These readable issues introduced by speakers and ASR systems will impair the performance of downstream tasks and the understanding of human readers. In this work, we present a task called ASR post-processing for readability (APR) and formulate it as a sequence-to-sequence text generation problem. The APR task aims to transform the noisy ASR output into a readable text for humans and downstream tasks while maintaining the semantic meaning of speakers. We further study the APR task from the benchmark dataset, evaluation metrics, and baseline models: First, to address the lack of task-specific data, we propose a method to construct a dataset for the APR task by using the data collected for grammatical error correction. Second, we utilize metrics adapted or borrowed from similar tasks to evaluate model performance on the APR task. Lastly, we use several typical or adapted pre-trained models as the baseline models for the APR task. Furthermore, we fine-tune the baseline models on the constructed dataset and compare their performance with a traditional pipeline method in terms of proposed evaluation metrics. Experimental results show that all the fine-tuned baseline models perform better than the traditional pipeline method, and our adapted RoBERTa model outperforms the pipeline method by 4.95 and 6.63 BLEU points on two test sets, respectively. The human evaluation and case study further reveal the ability of the proposed model to improve the readability of ASR transcripts.

Automatic Fluency Assessment Method for Spontaneous Speech without Reference Text

The automatic fluency assessment of spontaneous speech without reference text is a challenging task that heavily depends on the accuracy of automatic speech recognition (ASR). Considering this scenario, it is necessary to explore an assessment method that combines ASR. This is mainly due to the fact that in addition to acoustic features being essential for assessment, the text features output by ASR may also contain potentially fluency information. However, most existing studies on automatic fluency assessment of spontaneous speech are based solely on audio features, without utilizing textual information, which may lead to a limited understanding of fluency features. To address this, we propose a multimodal automatic speech fluency assessment method that combines ASR output. Specifically, we first explore the relevance of the fluency assessment task to the ASR task and fine-tune the Wav2Vec2.0 model using multi-task learning to jointly optimize the ASR task and fluency assessment task, resulting in both the fluency assessment results and the ASR output. Then, the text features and audio features obtained from the fine-tuned model are fed into the multimodal fluency assessment model, using attention mechanisms to obtain more reliable assessment results. Finally, experiments on the PSCPSF and Speechocean762 dataset suggest that our proposed method performs well in different assessment scenarios.

Real-time Context-Aware Multimodal Network for Activity and Activity-Stage Recognition from Team Communication in Dynamic Clinical Settings.

In clinical settings, most automatic recognition systems use visual or sensory data to recognize activities. These systems cannot recognize activities that rely on verbal assessment, lack visual cues, or do not use medical devices. We examined speech-based activity and activity-stage recognition in a clinical domain, making the following contributions. (1) We collected a high-quality dataset representing common activities and activity stages during actual trauma resuscitation events-the initial evaluation and treatment of critically injured patients. (2) We introduced a novel multimodal network based on audio signal and a set of keywords that does not require a high-performing automatic speech recognition (ASR) engine. (3) We designed novel contextual modules to capture dynamic dependencies in team conversations about activities and stages during a complex workflow. (4) We introduced a data augmentation method, which simulates team communication by combining selected utterances and their audio clips, and showed that this method contributed to performance improvement in our data-limited scenario. In offline experiments, our proposed context-aware multimodal model achieved F1-scores of 73.2±0.8% and 78.1±1.1% for activity and activity-stage recognition, respectively. In online experiments, the performance declined about 10% for both recognition types when using utterance-level segmentation of the ASR output. The performance declined about 15% when we omitted the utterance-level segmentation. Our experiments showed the feasibility of speech-based activity and activity-stage recognition during dynamic clinical events.

Streaming Punctuation: A Novel Punctuation Technique Leveraging Bidirectional Context for Continuous Speech Recognition

While speech recognition Word Error Rate (WER) has reached human parity for English, continuous speech recognition scenarios such as voice typing and meeting transcriptions still suffer from segmentation and punctuation problems, resulting from irregular pausing patterns or slow speakers. Transformer sequence tagging models are effective at capturing long bi-directional context, which is crucial for automatic punctuation. Automatic Speech Recognition (ASR) production systems, however, are constrained by real-time requirements, making it hard to incorporate the right context when making punctuation decisions. Context within the segments produced by ASR decoders can be helpful but limiting in overall punctuation performance for a continuous speech session. In this paper, we propose a streaming approach for punctuation or re-punctuation of ASR output using dynamic decoding windows and measure its impact on punctuation and segmentation accuracy across scenarios. The new system tackles over-segmentation issues, improving segmentation F0.5-score by 13.9%. Streaming punctuation achieves an average BLEUscore improvement of 0.66 for the downstream task of Machine Translation (MT).

Confusion2Vec 2.0: Enriching ambiguous spoken language representations with subwords.

Word vector representations enable machines to encode human language for spoken language understanding and processing. Confusion2vec, motivated from human speech production and perception, is a word vector representation which encodes ambiguities present in human spoken language in addition to semantics and syntactic information. Confusion2vec provides a robust spoken language representation by considering inherent human language ambiguities. In this paper, we propose a novel word vector space estimation by unsupervised learning on lattices output by an automatic speech recognition (ASR) system. We encode each word in Confusion2vec vector space by its constituent subword character n-grams. We show that the subword encoding helps better represent the acoustic perceptual ambiguities in human spoken language via information modeled on lattice-structured ASR output. The usefulness of the proposed Confusion2vec representation is evaluated using analogy and word similarity tasks designed for assessing semantic, syntactic and acoustic word relations. We also show the benefits of subword modeling for acoustic ambiguity representation on the task of spoken language intent detection. The results significantly outperform existing word vector representations when evaluated on erroneous ASR outputs, providing improvements up-to 13.12% relative to previous state-of-the-art in intent detection on ATIS benchmark dataset. We demonstrate that Confusion2vec subword modeling eliminates the need for retraining/adapting the natural language understanding models on ASR transcripts.

Automatic Speech Recognition Post-Processing for Readability: Task, Dataset and a Two-Stage Pre-Trained Approach

Nowadays Automatic Speech Recognition (ASR) systems can accurately recognize which words are said. However, due to the disfluency, grammatical error, and other phenomena in spontaneous speech, the verbatim transcription of ASR impairs its readability, which is crucial for human comprehension and downstream tasks processing that need to understand the meaning and purpose of what is spoken. In this work, we formulate the ASR post-processing for readability (APR) as a sequence-to-sequence text generation problem that aims to transform the incorrect and noisy ASR output into readable text for humans and downstream tasks. We leverage the Metadata Extraction (MDE) corpus to construct a task-specific dataset for our study. To solve the problem of too little training data, we propose a novel data augmentation method that synthesizes large-scale training data from the grammatical error correction dataset. We propose a model based on the pre-trained language model to perform the APR task and train the model with a two-stage training strategy to better exploit the augmented data. On the constructed test set, our approach outperforms the best baseline system by a large margin of 17.53 on BLEU and 13.26 on readability-aware WER (RA-WER). The human evaluation also shows that our model can generate more human-readable transcripts than the baseline method.

Streaming cascade-based speech translation leveraged by a direct segmentation model

The cascade approach to Speech Translation (ST) is based on a pipeline that concatenates an Automatic Speech Recognition (ASR) system followed by a Machine Translation (MT) system. Nowadays, state-of-the-art ST systems are populated with deep neural networks that are conceived to work in an offline setup in which the audio input to be translated is fully available in advance. However, a streaming setup defines a completely different picture, in which an unbounded audio input gradually becomes available and at the same time the translation needs to be generated under real-time constraints. In this work, we present a state-of-the-art streaming ST system in which neural-based models integrated in the ASR and MT components are carefully adapted in terms of their training and decoding procedures in order to run under a streaming setup. In addition, a direct segmentation model that adapts the continuous ASR output to the capacity of simultaneous MT systems trained at the sentence level is introduced to guarantee low latency while preserving the translation quality of the complete ST system. The resulting ST system is thoroughly evaluated on the real-life streaming Europarl-ST benchmark to gauge the trade-off between quality and latency for each component individually as well as for the complete ST system.

Automatic detection of prosodic boundaries in spontaneous speech.

Automatic speech recognition (ASR) and natural language processing (NLP) are expected to benefit from an effective, simple, and reliable method to automatically parse conversational speech. The ability to parse conversational speech depends crucially on the ability to identify boundaries between prosodic phrases. This is done naturally by the human ear, yet has proved surprisingly difficult to achieve reliably and simply in an automatic manner. Efforts to date have focused on detecting phrase boundaries using a variety of linguistic and acoustic cues. We propose a method which does not require model training and utilizes two prosodic cues that are based on ASR output. Boundaries are identified using discontinuities in speech rate (pre-boundary lengthening and phrase-initial acceleration) and silent pauses. The resulting phrases preserve syntactic validity, exhibit pitch reset, and compare well with manual tagging of prosodic boundaries. Collectively, our findings support the notion of prosodic phrases that represent coherent patterns across textual and acoustic parameters.

Neural candidate-aware language models for speech recognition

This paper presents novel neural network based language models that can correct automatic speech recognition (ASR) errors by using speech recognizer outputs as a context. Our proposed models, called neural candidate-aware language models (NCALMs), estimate the generative probability of a target sentence while considering ASR outputs including hypotheses and their posterior probabilities. Recently, neural network language models have achieved great success in ASR field because of their ability to learn long-range contexts and model the word representation in continuous space. However, they estimate a sentence probability without considering other candidates and their posterior probabilities, even though the competing hypotheses are available and include important information to increase the speech recognition accuracy. To overcome this limitation, our idea is to utilize ASR outputs in both the training phase and the inference phase. Our proposed models are conditional generative models consisting of a Transformer encoder and a Transformer decoder. The encoder embeds the candidates as context vectors and the decoder estimates a sentence probability given the context vectors. We evaluate the proposed models in Japanese lecture transcription and English conversational speech recognition tasks. Experimental results show that a NCALM has better ASR performance than a system including a deep neural network-hidden Markov model hybrid system. We further improve ASR performance by using a NCALM and a Transformer language model simultaneously.

An ensemble technique for speech recognition in noisy environments

<span>Automatic speech recognition (ASR) is a technology that allows a computer and mobile device to recognize and translate spoken language into text. ASR systems often produce poor accuracy for the noisy speech signal. Therefore, this research proposed an ensemble technique that does not rely on a single filter for perfect noise reduction but incorporates information from multiple noise reduction filters to improve the final ASR accuracy. The main factor of this technique is the generation of K-copies of the speech signal using three noise reduction filters. The speech features of these copies differ slightly in order to extract different texts from them when processed by the ASR system. Thus, the best among these texts can be elected as final ASR output. The ensemble technique was compared with three related current noise reduction techniques in terms of CER and WER. The test results were encouraging and showed a relatively decreased by 16.61% and 11.54% on CER and WER compared with the best current technique. ASR field will benefit from the contribution of this research to increase the recognition accuracy of a human speech in the presence of background noise.</span>

A low latency sequential model and its user-focused evaluation for automatic punctuation of ASR closed captions

In Automatic Speech Recognition (ASR), inserting the punctuation marks into the word chain hypothesis has long been given low priority, as efforts were concentrated on minimizing word error rates. Punctuation, however, also has a high impact on the transcription quality perceived by the users. Prosody, textual context and their combination have since been used successfully for automatic punctuation of ASR outputs. The recently proposed RNN based solutions show encouraging performance. We believe that current bottlenecks of punctuation technology are on one hand the complex punctuation models, which, having high latency, are not suitable for use-cases with real-time requirements; and on the other hand, punctuation efforts have not been validated against human perception and user impression. The ambition of this paper is to propose a lightweight, yet powerful RNN punctuation model for on-line (real-time including low latency) environment, and also to assess user opinion, in general and also for target users living with hearing loss or impairment. The proposed on-line RNN punctuation model is evaluated against a Maximum Entropy (MaxEnt) baseline, for Hungarian and for English, whereas subjective assessment tests are carried out on real broadcast data subtitled with ASR (closed captioning). As it can be expected, the RNN outperforms the MaxEnt baseline system, but of course not the off-line systems: limiting the future context to minimize latency results only in a slighter performance drop, but ASR errors obviously influence punctuation performance considerably. A genre analysis is also carried out w.r.t. the punctuation performance showing that both recognition and punctuation of more spontaneous speech styles is challenging. Overall, the subjective tests confirmed that users perceive a significant quality improvement when punctuation is added, even in presence of word errors and even if punctuation is automatic and hence itself may contain further errors. For users living with hearing loss or deafness, an even higher, clear preference for the punctuated captions could be confirmed.

Sequence labeling to detect stuttering events in read speech

Stuttering is a speech disorder that, if treated during childhood, may be prevented from persisting into adolescence. A clinician must first determine the severity of stuttering, assessing a child during a conversational or reading task, recording each instance of disfluency, either in real time, or after transcribing the recorded session and analysing the transcript. The current study evaluates the ability of two machine learning approaches, namely conditional random fields (CRF) and bi-directional long-short-term memory (BLSTM), to detect stuttering events in transcriptions of stuttering speech. The two approaches are compared for their performance both on ideal hand-transcribed data and also on the output of automatic speech recognition (ASR). We also study the effect of data augmentation to improve performance. A corpus of 35 speakers’ read speech (13K words) was supplemented with a corpus of 63 speakers’ spontaneous speech (11K words) and an artificially-generated corpus (50K words). Experimental results show that, without feature engineering, BLSTM classifiers outperform CRF classifiers by 33.6%. However, adding features to support the CRF classifier yields performance improvements of 45% and 18% over the CRF baseline and BLSTM results, respectively. Moreover, adding more data to train the CRF and BLSTM classifiers consistently improves the results.

Application of automatic speech recognition (ASR) techniques for automatic speech assessment in people with aphasia

Application of automatic speech recognition (ASR) techniques for automatic speech assessment in people with aphasia

Using automatic speech recognition to assess spoken responses to cognitive tests of semantic verbal fluency

Cognitive tests of verbal fluency (VF) consist of verbalizing as many words as possible in one minute that either start with a specific letter of the alphabet or belong to a specific semantic category. These tests are widely used in neurological, psychiatric, mental health, and school settings and their validity for clinical applications has been extensively demonstrated. However, VF tests are currently administered and scored manually making them too cumbersome to use, particularly for longitudinal cognitive monitoring in large populations. The objective of the current study was to determine if automatic speech recognition (ASR) could be used for computerized administration and scoring of VF tests. We examined established techniques for constraining language modeling to a predefined vocabulary from a specific semantic category (e.g., animals). We also experimented with post-processing ASR output with confidence scoring, as well as with using speaker adaptation to improve automated VF scoring. Audio responses to a VF task were collected from 38 novice and experienced professional fighters (boxing and mixed martial arts) participating in a longitudinal study of effects of repetitive head trauma on brain function. Word error rate, correlation with manual word count and distance from manual word count were used to compare ASR-based approaches to scoring to each other and to the manually scored reference standard. Our study’s results show that responses to the VF task contain a large number of extraneous utterances and noise that lead to relatively poor baseline ASR performance. However, we also found that speaker adaptation combined with confidence scoring significantly improves all three metrics and can enable use of ASR for reliable estimates of the traditional manual VF scores.

Spoken Content Retrieval—Beyond Cascading Speech Recognition with Text Retrieval

Spoken content retrieval refers to directly indexing and retrieving spoken content based on the audio rather than text descriptions. This potentially eliminates the requirement of producing text descriptions for multimedia content for indexing and retrieval purposes, and is able to precisely locate the exact time the desired information appears in the multimedia. Spoken content retrieval has been very successfully achieved with the basic approach of cascading automatic speech recognition (ASR) with text information retrieval: after the spoken content is transcribed into text or lattice format, a text retrieval engine searches over the ASR output to find desired information. This framework works well when the ASR accuracy is relatively high, but becomes less adequate when more challenging real-world scenarios are considered, since retrieval performance depends heavily on ASR accuracy. This challenge leads to the emergence of another approach to spoken content retrieval: to go beyond the basic framework of cascading ASR with text retrieval in order to have retrieval performances that are less dependent on ASR accuracy. This overview article is intended to provide a thorough overview of the concepts, principles, approaches, and achievements of major technical contributions along this line of investigation. This includes five major directions: 1) Modified ASR for Retrieval Purposes: cascading ASR with text retrieval, but the ASR is modified or optimized for spoken content retrieval purposes; 2) Exploiting the Information not present in ASR outputs: to try to utilize the information in speech signals inevitably lost when transcribed into phonemes and words; 3) Directly Matching at the Acoustic Level without ASR: for spoken queries, the signals can be directly matched at the acoustic level, rather than at the phoneme or word levels, bypassing all ASR issues; 4) Semantic Retrieval of Spoken Content: trying to retrieve spoken content that is semantically related to the query, but not necessarily including the query terms themselves; 5) Interactive Retrieval and Efficient Presentation of the Retrieved Objects: with efficient presentation of the retrieved objects, an interactive retrieval process incorporating user actions may produce better retrieval results and user experiences.

Level of interest sensing in spoken dialog using decision-level fusion of acoustic and lexical evidence

Automatic detection of a user's interest in spoken dialog plays an important role in many applications, such as tutoring systems and customer service systems. In this study, we propose a decision-level fusion approach using acoustic and lexical information to accurately sense a user's interest at the utterance level. Our system consists of three parts: acoustic/prosodic model, lexical model, and a model that combines their decisions for the final output. We use two different regression algorithms to complement each other for the acoustic model. For lexical information, in addition to the bag-of-words model, we propose new features including a level-of-interest value for each word, length information using the number of words, estimated speaking rate, silence in the utterance, and similarity with other utterances. We also investigate the effectiveness of using more automatic speech recognition (ASR) hypotheses (n-best lists) to extract lexical features. The outputs from the acoustic and lexical models are combined at the decision level. Our experiments show that combining acoustic evidence with lexical information improves level-of-interest detection performance, even when lexical features are extracted from ASR output with high word error rate.