Rohit Sinha

Mithibai CollegeUniversity of Mumbai, Bachelor Of Arts (B.A), Undergraduate

Followers

Following

Public Views

Interests

Uploads

Papers by Rohit Sinha

Semi-Coupled Dictionary Based Automatic Bandwidth Extension Approach for Enhancing Children’s ASR

Interspeech 2016, 2016

The work presented in this paper is motivated by our earlier work exploring sparse representation... more The work presented in this paper is motivated by our earlier work exploring sparse representation based approach for automatic bandwidth extension (ABWE) of speech signals. In that work, two dictionaries one for voiced and the other for unvoiced speech frames are created using KSVD algorithm on wideband data. Each of the atoms of these dictionaries is then decimated and interpolated by a factor of 2 to generate narrowband interpolated (NBI) dictionaries whose atoms have one-to-one correspondence with those of the WB dictionaries. The given narrowband speech frames are also interpolated to generated NBI targets and those are sparse coded over the NBI dictionaries. The resulting sparse codes are then applied to the WB dictionaries to estimate the WB target data. In this work, we extend the said approach by making use of an existing semi-coupled dictionary learning (SCDL) algorithm. Unlike the direct dictionary learning, the SCDL algorithm also learns a set of bidirectional transforms coupling the dictionaries more flexibly. The bandwidth enhanced speech obtained employing the SCDL approach and a modified high/low band gain adjustment yields significant improvements in terms of speech quality measures as well as in the context of children's mismatched speech recognition.

Download

A Novel Approach for Effective Recognition of the Code-Switched Data on Monolingual Language Model

Interspeech 2018, 2018

Code-switching refers to the phenomena of mixing of words or phrases from foreign languages while... more Code-switching refers to the phenomena of mixing of words or phrases from foreign languages while communicating in a native language by the multilingual speakers. Codeswitching is a global phenomenon and is widely accepted in multilingual communities. However, for training the language model (LM) for such tasks, a very limited code-switched textual resources are available as yet. In this work, we present an approach to reduce the perplexity (PPL) of Hindi-English code-switched data when tested over the LM trained on purely native Hindi data. For this purpose, we propose a novel textual feature which allows the LM to predict the code-switching instances. The proposed feature is referred to as code-switching factor (CS-factor). Also, we developed a tagger that facilitates the automatic tagging of the code-switching instances. This tagger is trained on a development data and assigns an equivalent class of foreign (English) words to each of the potential native (Hindi) words. For this study, the textual resource has been created by crawling the blogs from a couple of websites educating about the usage of the Internet. In the context of recognition of the code-switching data, the proposed technique is found to yield a substantial improvement in terms of PPL.

Download

A simple approach to non-uniform vowel normalization

IEEE International Conference on Acoustics Speech and Signal Processing, 2002

Speech biometric based attendance system

2014 Twentieth National Conference on Communications (NCC), 2014

Speech based Emotion Recognition based on hierarchical decision tree with SVM, BLG and SVR classifiers

2013 National Conference on Communications (NCC), 2013

ABSTRACT Emotion Recognition is increasingly becoming an important part of computer vision and ro... more ABSTRACT Emotion Recognition is increasingly becoming an important part of computer vision and robotics. There has been a lot of research and development around this field in the recent times. It is imperative to design emotion recognition systems for real time situations having a considerable rate of accuracy which can find its application in telecommunications, security, etc. This paper discusses a novel design/approach based on a hierarchical decision tree for the GMM means supervector based feature set and using various classifiers viz. SVM, BLG and SVR, to improve the performance of the existing emotion recognition systems. These approaches have been studied on Emo-DB, a German language emotional speech database, yielding about 83% recognition results for closed set based speaker-independent recognition for the optimized method. These results are similar to the results achieved by existing studies on emotion recognition using GMM means supervectors.

Modified Joint Channel Source Decoding structure for reliable image transmission

2012 International Conference on Communication, Information & Computing Technology (ICCICT), 2012

ABSTRACT Transmission of images over unreliable channels is highly prone to errors. The existing ... more ABSTRACT Transmission of images over unreliable channels is highly prone to errors. The existing Turbo code based Joint Channel Source Decoding (JCSD) uses weighting operation on the extrinsic information passed between the maximum a posteriori probability (MAP) decoders to minimize the transmission errors. We propose a modification to the existing structure to further enhance the visual quality by minimizing the transmission errors. In the modified structure, the weighting operation is applied on the channel soft output values instead of extrinsic information. Also, a low complexity error detection algorithm based on variance/ standard deviation and channel soft output value is proposed. Simulation results show the Peak Signal to Noise Ratio (PSNR) improvement of about 0.8-1.0 dB over existing JCSD approach for a low channel SNR value of 1dB.

Soft-weighting technique for robust children speech recognition under mismatched condition

2013 Annual IEEE India Conference (INDICON), 2013

ABSTRACT The children&#39;s speech recognition performance under mismatched condition i.e., r... more ABSTRACT The children&#39;s speech recognition performance under mismatched condition i.e., recognizing on the adults&#39; speech trained models is a challenging task. It is well known that MFCC features contain all the information regarding speech and mismatch factor at the same time. Therefore, in this work, the truncation of MFCC features is explored for children&#39;s speech recognition on adults&#39; speech trained models. It has already been noted that cepstral truncation gives improved result but excessive increase in cepstral truncation will cause the loss of the relevant spectral information. Motivated by this, in this work, we explored soft-weighing technique that is heteroscedastic linear discriminant analysis (HLDA) for optimizing losses of cepstral information during truncation. In this paper, an HLDA transformation based technique to reduce mismatch condition is proposed. Further, we have tried to develop a linear relationship between HLDA transformation subspace of MFCC features and the VTLN warp factor values. Finally, a scheme to concatenate VTLN and CMLLR is explored. The proposed approach was found to give improvement in performance by 42.18% and 18.88% in the case of connected digit recognition and continuous speech recognition respectively in comparison to direct cepstral truncation.

Robustness of speaker normalization approaches: A study

TENCON 2008 - 2008 IEEE Region 10 Conference, 2008

Abstract In this work, we have studied the effectiveness of recently proposed warp factor estimat... more

Image denoising using ridgelet transform in a collaborative filtering framework

2014 International Conference on Signal Processing and Communications (SPCOM), 2014

Redundancy within a natural image has been widely exploited for enhancing the sparsity in represe... more Redundancy within a natural image has been widely exploited for enhancing the sparsity in representation to achieve better restoration. The BM3D is the current state-of-the-art denoising method. It combines the transform-domain and the non-local means spatial-domain algorithms and is named as “collaborative filtering”. The discrete wavelet transform is shown to be the best performing transform in the collaborative filtering framework for Gaussian noise removal from the natural images. The BM3D is based on the assumption of homogeneity. However, in case of non-homogeneous textured image, there are structures such as lines, curves, etc. which still need to be exploited carefully. Due to the inherent limitations of the wavelets in representing the line singularity, the ridgelet transform was proposed a decade ago. In this paper, we explore the use of ridgelet transform in collaborative filtering framework and show that the BM3D performs better with the ridgelet transform than with the wavelet transform for the denoising textured image.

A shift-based approach to speaker normalization using non-linear frequency-scaling model

Speech Communication, 2008

In this work, we present a speaker-normalization method based on the idea that the speaker-depend... more In this work, we present a speaker-normalization method based on the idea that the speaker-dependent scale-factor can be separated out as a fixed translation factor in an alternate domain. We also introduce a non-linear frequency-scaling model motivated by the analysis of speech data. The proposed shift-based normalization approach is implemented using a maximum-likelihood (ML) search for the translation factor in the alternate domain. The advantage of our approach is that we are able to show the relationship between conventional frequency-warping based vocal-tract length normalization (VTLN) methods and the methods based on shifts in psycho-acoustic scale thus providing a unifying framework for speaker-normalization. Additionally, in our approach it is simple to show that the shifting required for normalization can be expressed as a linear transformation in the cepstral domain. This is important for computational efficiency since we do not have to recompute the features by redoing the signal processing for each scale/translation factor as is usually done in conventional normalization. We present recognition results using our proposed approach on a digit recognition task and show that the non-linear scaling model provides relative improvement of 4% for adults and 7.5% for children when compared to the linear-scaling model.

Download

Nonuniform speaker normalization using affine transformation

The Journal of the Acoustical Society of America, 2008

In this paper, we propose a mathematical model to describe the relation between the formant frequ... more In this paper, we propose a mathematical model to describe the relation between the formant frequencies of speakers and show that with the proposed affine model, speaker differences separate out as translation factors when a "mel-like" warping is performed. Using speech data we estimate the parameters of this warping function and show that it is close to the usual mel-formula. This model is motivated by Rohit et al.'s [1] shift-based non-uniform speakernormalization method, which provides improvement over the conventional maximum-likelihood based speaker normalization methods. We therefore provide a unified framework that relates the relationship between formants of speakers and method of removing speakers difference (which involves mel-warping) in a neat mathematical framework which is substantiated by our recognition experiments.

Download

Multi-variability speech database for robust speaker recognition

National Conference on Communications, 2011

In this paper, we present our initial study with the recently collected speech database for devel... more In this paper, we present our initial study with the recently collected speech database for developing robust speaker recognition systems in Indian context. The database contains the speech data collected across different sensors, languages, speaking styles, and environments, from 200 speakers. The speech data is collected across five different sensors in parallel, in English and multiple Indian languages, in reading

Multivariability Speaker Recognition Database in Indian Scenario

jeeonline.iitg.ernet.in

In this paper we describe the collection and organization of the speaker recognition database in ... more In this paper we describe the collection and organization of the speaker recognition database in Indian scenario named as IITG Multivariability Speaker Recognition Database. The database contains speech from 451 speakers speaking English and other Indian languages both in conversational and read speech styles recorded using various sensors in parallel under different environmental conditions. The database is organized into four phases on the basis of different conditions employed for the recording. The results of the initial studies conducted on a speaker verification system exploring the impact of mismatch in training and test conditions using the collected data are also included. A copy of this database can be obtained from the authors by contacting them.

Download

Exploring Data-Independent Dimensionality Reduction in Sparse Representation-Based Speaker Identification

Circuits, Systems, and Signal Processing, 2014

The sparse representation classification (SRC) has attracted the attention of many signal process... more The sparse representation classification (SRC) has attracted the attention of many signal processing domains in past few years. Recently, it has been successfully explored for the speaker recognition task with Gaussian mixture model (GMM) mean supervectors which are typically of the order of tens of thousands as speaker representations. As a result of this, the complexity of such systems become very high. With the use of the state-of-the-art i-vector representations, the dimension of GMM mean supervectors can be reduced effectively. But the i-vector approach involves a high dimensional data projection matrix which is learned using the factor analysis approach over huge amount of data from a large number of speakers. Also, the estimation of i-vector for a given utterance involves a computationally complex procedure. Motivated by these facts, we explore the use of data-independent projection approaches for reducing the dimensionality of GMM mean supervectors. The data-independent projection methods studied in this work include a normal random projection and two kinds of sparse random projections. The study is performed on SRC-based speaker identification using the NIST SRE 2005 dataset which includes channel matched and mismatched conditions. We find that the use of data-independent random projections for the dimensionality reduction of the supervectors results in only 3 % absolute loss in performance compared to that of the data-dependent (i-vector) approach. It is highlighted that with the use of highly sparse random projection matrices having $$\pm $$±1 as non-zero coefficients, a significant reduction in computational complexity is achieved in finding the projections. Further, as these matrices do not require floating point representations, their storage requirement is also very small compared to that of the data-dependent or the normal random projection matrices. These reduced complexity sparse random projections would be of interest in context of the speaker recognition applications implemented on platforms having low computational power.

A Study into Front-End Signal Processing for Automatic Speech Recognition

Workshop on Spoken Language Processing, 2003

In our receutly reported work, we have observed some difference in recognition performance using ... more

Exploring the effect of differences in the acoustic correlates of adults' and children's speech in the context of automatic speech recognition

EURASIP Journal on Audio, Speech, and Music …, 2010

This work explores the effect of mismatches between adults' and children's speech due t... more

Study Of Non-Linear Frequency Warping Functions For Speaker Normalization

Acoustics, Speech and Signal …

Special Section on Rich Transcription

Improving speech transcription for Mandarin-English translation

The development of the Cambridge University RT-04 diarisation system

Rohit Sinha

Uploads

Papers by Rohit Sinha

Log In