Table of contentsIntroductionMethodologyAutomatic speech recognition (ASR)Experiment and resultConclusion and discussionIntroductionSpeech recognition is the ability of a machine, to a database or program to recognize sentences, expressions or words in spoken language and translate them into a machine-readable layout. Detecting dialogue from non-native speakers is, in itself, a very challenging task. The speech around voice recognition has been around for years, but it's worth asking why it matters today. The motivation is that deep learning has finally made speech recognition accurate enough to be useful outside of a carefully controlled environment. Say no to plagiarism. Get a tailor-made essay on “Why Violent Video Games Should Not Be Banned”? Get an original essay. Machine learning is the realization that there are universal algorithms that can tell something fascinating and interesting about a set of data without you having to do it. enter any custom code specific to the problem. Instead of writing codes, data is entered into the generic algorithm and it builds its logic based on the data. Suffice it to mention that with the current arrangement of the world, the diversity of users nevertheless highlights serious considerations. However, the prerogative that all users should have equal access to voice recognition is not strong. This goes back to the analogy that people with pitiful reading skills don't have the same access to newspapers as highly educated people. Simply put, machine learning is an umbrella term that houses different types of generic algorithms. Additionally, non-native recognition plays a key role in border control security systems. Such recognition systems help security officials identify immigrants with counterfeit and falsified permits or IDs by pinpointing the country in which the foreign accent is pronounced. Additionally, it appears that voice recognition applications are poised to become a default interface for information delivery systems. Housing users whose language use is compromised in one way or another is not only a research problem but also a useful and crucial concern. Methodology There are, however, few considerations that seem particularly useful for coding non-native speech. Modes, choice, lexical and syntactic soundness, accent, and fluency are facets of spoken English that can both label disparity in native speech and be used to differentiate it from native speakers. "The accent generally comes from the speaker's habit of articulation in his language." As it is widely known that beginners in a language are mostly exposed to preliminary grammar at the premature stage of their study, imperfect command of syntax is one of the types that can make even a very proficient speech , non-native speech. Lately, the perception between native and non-native speech has been undermined using binary classification structures. These frameworks mainly depend on prosodic, cepstral, speech recognition or N-gram based language types, and use support vector machines (SVM) for classification. Automatic Speech Recognition (ASR) Efforts to build automatic speech recognition systemsSpeech recognition systems (ASR) were first carried out in the 1950s. These early speech recognition systems attempted to link together a set of grammatical and syntactic rules to identify speech. The system can only identify the word if the spoken words follow a certain set of rules. An automatic speech recognition (ASR) module forms the root of virtually all spoken language assessment systems. An ASR front-end component for most state-of-the-art assessment systems gives verbal speculations about the answers given by the person available for assessment. Therefore, one can predict that a huge amount of data, specifically a pool of non-native speech, and careful transcriptions of every piece of that speech, would be required to train this type of ASR module. Furthermore, it does not no doubt it would involve human effort to transcribe the entire collection of speeches. Despite advances in automatic speech recognition (ASR) systems that have led to supporters, developing robust ASR systems that provide high performance to diverse user groups remains a challenge. The problem with current ASR systems is that they primarily work with native speech. only, and accuracy drops significantly when words are articulated with an unusual pronunciation (foreign accent). However, human language makes many concessions to its guidelines. The way words and sentences are articulated can be changed enormously by dialects, consents and mannerisms. First, there is a disparity in what the speaker is saying. For open vocabulary systems, there is no way to collect training data for every imaginable utterance or even for every possible word. Second, there is dissimilarity due to differences between speakers. Different people have different voices, accents, and ways of speaking. Third, noise conditions vary. Anything in the acoustic data that is not the signal is noise, and noise can therefore include background sounds, microphone-specific artifacts, and other effects. Therefore, to achieve automatic speech recognition, we use deep learning algorithm. Therefore, for this study, deep learning algorithm will be considered as our methodology. It may also be interesting to know that deep learning researchers who know almost nothing about language translation are building relatively simple machine learning solutions that are now beating the best language translation systems designed by experts. Network (Deep Learning) is a construct particularly used for clustering or regression tasks when the extraordinary dimensionality and nonlinearity of the data make these tasks unlikely to be accomplished. In the case of visual data, the standard is to engage convolutional neural networks (CNN). CNNs are directly inspired by the hierarchy of cells in visual neuroscience. It is important to note that the neural network itself is not an algorithm, but rather a blueprint for many other machine learning algorithms to work collectively and process multiple and complex data inputs. Siniscalchi et al. , (2013) have already established that the manner and location of articulation attributes can effectively characterize any spoken language in the same way as in the Automatic Speech Attribute Transcription (ASAT) model for automatic recognition of speech. Keep in mind: this is just one..