Automatic spoken language identification using MFCC based time series features

The use of face masks has increased dramatically since the COVID-19 pandemic started in order to to curb the spread of the disease. Additionally, breakthrough infections caused by the Delta and Omicron variants have further increased the importance of wearing a face mask, even for vaccinated individuals. However, the use of face masks also induces attenuation in speech signals, and this change may impact speech processing technologies, e.g., automated speaker verification (ASV) and speech to text conversion. In this paper we examine Automatic Speaker Verification (ASV) systems against the speech samples in the presence of three different types of face mask: surgical, cloth, and filtered N95, and analyze the impact on acoustics and other factors. In addition, we explore the effect of different microphones, and distance from the microphone, and the impact of face masks when speakers use ASV systems in real-world scenarios. Our analysis shows a significant deterioration in performance when an ASV system encounters different face masks, microphones, and variable distance between the subject and microphone. To address this problem, this paper proposes a novel framework to overcome performance degradation in these scenarios by realigning the ASV system. The novelty of the proposed ASV framework is as follows: first, we propose a fused feature descriptor by concatenating the novel Ternary Deviated overlapping Patterns (TDoP), Mel Frequency Cepstral Coefficients (MFCC), and Gammatone Cepstral Coefficients (GTCC), which are used by both the ensemble learning-based ASV and anomaly detection system in the proposed ASV architecture. Second, this paper proposes an anomaly detection model for identifying vocal samples produced in the presence of face masks. Next, it presents a Peak Norm (PN) filter to approximate the signal of the speaker without a face mask in order to boost the accuracy of ASV systems. Finally, the features of filtered samples utilizing the PN filter and samples without face masks are passed to the proposed ASV to test for improved accuracy. The proposed ASV system achieved an accuracy of 0.99 and 0.92, respectively, on samples recorded without a face mask and with different face masks. Although the use of face masks affects the ASV system, the PN filtering solution overcomes this deficiency up to 4%. Similarly, when exposed to different microphones and distances, the PN approach enhanced system accuracy by up to 7% and 9%, respectively. The results demonstrate the effectiveness of the presented framework against an in-house prepared, diverse Multi Speaker Face Masks (MSFM) dataset, (IRB No. FY2021-83), consisting of samples of subjects taken with a variety of face masks and microphones, and from different distances.

/pdf/toward-realigning-automatic-speaker-verification-in-the-era-1g21rh24.pdf

Toward Realigning Automatic Speaker Verification in the Era of COVID-19

System of methods of automated cognitive linguistic analysis of speech signals with noise

Automatic spoken language identification (LID) is a very important research field in the era of multilingual voice-command-based human-computer interaction (HCI). A front-end LID module helps to improve the performance of many speech-based applications in the multilingual scenario. India is a populous country with diverse cultures and languages. The majority of the Indian population needs to use their respective native languages for verbal interaction with machines. Therefore, the development of efficient Indian spoken language recognition systems is useful for adapting smart technologies in every section of Indian society. The field of Indian LID has started gaining momentum in the last two decades, mainly due to the development of several standard multilingual speech corpora for the Indian languages. Even though significant research progress has already been made in this field, to the best of our knowledge, there are not many attempts to analytically review them collectively. In this work, we have conducted one of the very first attempts to present a comprehensive review of the Indian spoken language recognition research field. In-depth analysis has been presented to emphasize the unique challenges of low-resource and mutual influences for developing LID systems in the Indian contexts. Several essential aspects of the Indian LID research, such as the detailed description of the available speech corpora, the major research contributions, including the earlier attempts based on statistical modeling to the recent approaches based on different neural network architectures, and the future research trends are discussed. This review work will help assess the state of the present Indian LID research by any active researcher or any research enthusiasts from related fields.

/pdf/an-overview-of-indian-spoken-language-recognition-from-1ol0gl7o.pdf

An Overview of Indian Spoken Language Recognition from Machine Learning Perspective

Anti-spoofing research plays an important role in audio forensics. It has found a lot of traction in several languages around the world. With that in mind, the purpose of this work is to assess the impact of several synthetic spoofing detection approaches on a multilingual, low-constrained Indian language set. This paper aims at a multitasking spoofing detection by identifying real/spoof utterance identification as well as the regional language spoofing attack vector. To accomplish this, the features and the classifiers that are best candidate for the synthetic spoofing detection and language identification are appropriately chosen. Our methodology compares the performances of three main different classifiers GMM, SVM, DNN on the vector formulated from the accumulation of MFCC features. Hindi, Malayalam, Tamil, Telugu are the four languages which are taken into account for the comparison. Out of these classifiers, SVM and DNN are found to give the best results with EER rates of 1.98 % and 1.19 % respectively. 

Multi Tasking Synthetic Speech Detection on Indian Languages

Spoken language has been the most natural mode of communication since a long time. In today’s world, the advancement of telecommunication technology has taken communication to the next level. This acts as motivation for an automatic speech recognition (ASR) process. As research in ASR progresses, the development of language identification would be necessary for any multi-lingual ASR system. This problem is highly pertinent to India due to the existence of a wide diversity of languages, each having hundreds of dialects. In this chapter, we propose a spoken language identification framework for the recognition of the six most widely used spoken languages in India namely, English, Hindi, Bangla, Marathi, Tamil and Telugu. Firstly, the pre-processing is performed on an input audio clip, which focuses on removing its silent frames. This, in turn, increases its probability of recognition as redundant and non-informative information gets eliminated. The silence removal parameters are optimized in order to achieve the best performance. Secondly, the Mel Frequency Cepstral Coefficient (MFCC) features are extracted from the speech signals. Using these features, a Support Vector Machine (SVM) classifier is trained, which has been established as a superior classifier, due to its large-margin classification property, and its ability to classify complex and non-linear data. The proposed framework is tested on a standard IndicTTS speech database. SVM is trained for 13 (static), 26 (static + delta) and 39 (static + delta + delta-delta) MFCC features. It is found that the best results are achieved using only 13 static features, and addition of delta and delta-delta features decreases the performance. Maximum accuracy obtained is 60.33% for the above-mentioned 6 languages, and 89.33% for only Bangla, English and Hindi.

Spoken Language Identification of Indian Languages Using MFCC Features

Formation of arbitrary shapes and patterns using swarm robots is a fairly well researched topic. However, there is still a long way to go. This is due to the hindrance caused by the immense complexity of distributed algorithms and their lack of scalability. This paper proposes a scalable algorithm that allows a set of identical, homogeneous, disk shaped fat mobile robots to form multiple uniform circles. All the robots run the same proposed (completely distributed and deterministic) algorithm. The robots follow a global coordinate system. All the points that the robots must occupy are pre-computed by them. The robots in this paper are not oblivious, and use a message passing technique to prevent collisions. The robots have the same limited visibility, within which they can observe other robots. Based on one’s current position and position of other robots in its visibility, a robot takes appropriate decisions so that the objective can be fulfilled. The algorithm guarantees formation of multiple uniform circles under a synchronous environment. At the end of the paper, we also discuss why this algorithm is independent of the shape the robots have to form, and can be extended to the formation of any arbitrary geometric shape by just changing the computation of the points that the robots eventually occupy.

Saif Rahaman

Papers

Automatic spoken language identification using MFCC based time series features

Spoken Language Identification of Indian Languages Using MFCC Features

Multiple Uniform Circle Formation by Fat Robots Under Limited Visibility

Uniform Distribution of Fat Robots on a Circle Under Limited Visibility