Ajay Srinivasamurthy
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Ajay Srinivasamurthy has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Xavier Serra
Universitat Pompeu Fabra, Barcelona, Spain
4 shared publications Andre Holzapfel
Music Technology Group, Universitat Pompeu Fabra, Barcelona, Spain
2 shared publications- Hartmut Helmke
German Aerospace Center (DLR), Institute of Flight Guidance, Lilienthalplatz 7, 38108 Braunschweig, Germany
2 shared publications - Petr Motlicek
Idiap Research Institute
2 shared publications - Youssef Oualil
Apple USA
2 shared publications Ali Taylan Cemgil
Dept. of Computer Engineering, Bogazici University, Istanbul, Turkey
1 shared publication
Latest Publications
- Open Access
Unified Modeling of Multi-Domain Multi-Device ASR Systems
Modern Automatic Speech Recognition (ASR) systems often use a portfolio of domain-specific models in order to get high accuracy for distinct user utterance types across different devices. In this paper, we propose an innovative approach that integrates the different per-domain per-device models into a unified model, using a combination of domain embedding, domain experts, mixture of experts and adversarial training. We run careful ablation studies to show the benefit of each of these innovations in contributing to the accuracy of the overall unified model. Experiments show that our proposed unified modeling approach actually outperforms the carefully tuned per-domain models, giving relative gains of up to 10% over a baseline model with negligible increase in the number of parameters.
Open AccessUnified Modeling of Multi-Domain Multi-Device ASR Systems
Modern Automatic Speech Recognition (ASR) systems often use a portfolio of domain-specific models in order to get high accuracy for distinct user utterance types across different devices. In this paper, we propose an innovative approach that integrates the different per-domain per-device models into a unified model, using a combination of domain embedding, domain experts, mixture of experts and adversarial training. We run careful ablation studies to show the benefit of each of these innovations in contributing to the accuracy of the overall unified model. Experiments show that our proposed unified modeling approach actually outperforms the carefully tuned per-domain models, giving relative gains of up to 10% over a baseline model with negligible increase in the number of parameters.
- Open Access
- 2 September 2018
- Published by International Speech Communication Association
- p. 3519-3523
Iterative Learning of Speech Recognition Models for Air Traffic Control
- 2 September 2018
- Published by International Speech Communication Association
- p. 3519-3523
Open AccessIterative Learning of Speech Recognition Models for Air Traffic Control
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 404-408
A context-aware speech recognition and understanding system for air traffic control domain
Automatic Speech Recognition and Understanding (ASRU) systems can generally use temporal and situational context information to improve their performance for a given task. This is typically done by rescoring the ASR hypotheses or by dynamically adapting the ASR models. For some domains, such as Air Traffic Control (ATC), this context information can be, however, small in size, partial and available only as abstract concepts (e.g. airline codes), which are difficult to map into full possible spoken sentences to perform rescoring or adaptation. This paper presents a multi-modal ASRU system, which dynamically integrates partial temporal and situational ATC context information to improve its performance. This is done either by 1) extracting word sequences which carry relevant ATC information from ASR N-best Lists and then perform a context-based rescoring on the extracted ATC segments or 2) by a partial adaptation of the language model. Experiments conducted on 4 hours of test data from Prague and Vienna approach (arrivals) showed a relative reduction of the ATC command error rate metric by 30% to 50%.
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 404-408
A context-aware speech recognition and understanding system for air traffic control domain
Automatic Speech Recognition and Understanding (ASRU) systems can generally use temporal and situational context information to improve their performance for a given task. This is typically done by rescoring the ASR hypotheses or by dynamically adapting the ASR models. For some domains, such as Air Traffic Control (ATC), this context information can be, however, small in size, partial and available only as abstract concepts (e.g. airline codes), which are difficult to map into full possible spoken sentences to perform rescoring or adaptation. This paper presents a multi-modal ASRU system, which dynamically integrates partial temporal and situational ATC context information to improve its performance. This is done either by 1) extracting word sequences which carry relevant ATC information from ASR N-best Lists and then perform a context-based rescoring on the extracted ATC segments or 2) by a partial adaptation of the language model. Experiments conducted on 4 hours of test data from Prague and Vienna approach (arrivals) showed a relative reduction of the ATC command error rate metric by 30% to 50%.
- Open Access
- 31 October 2017
- Published by Frontiers Media SA in Frontiers in Digital Humanities
- Vol. 4
Aspects of Tempo and Rhythmic Elaboration in Hindustani Music: A Corpus Study
This paper provides insights into aspects of tempo and rhythmic elaboration in Hindustani music, based on a study of a large corpus of recorded performances. Typical tempo developments and stress patterns within a metrical cycle are computed, which we refer to as tempo and rhythm patterns, respectively. Rhythm patterns are obtained by aggregating spectral features over metrical cycles. They reflect percussion patterns that are frequent in the corpus, and enable a discussion of the relation between such patterns and the underlying metrical framework, the taal. Tempo patterns, on the other hand, are computed using reference beat annotations. They document the dynamic development of tempo throughout a metrical cycle, and reveal insights into the flexibility of time in Hindustani music for the first time using quantitative methods on a large set of performances. Focusing on aspects of tempo and rhythm, we demonstrate the value of a computational methodology for the analysis of large music corpora by revealing the range of tempi used in performances, intra-cycle tempo dynamics and percussion accents at different positions of the taal cycle.
- 31 October 2017
- Published by Frontiers Media SA in Frontiers in Digital Humanities
- Vol. 4
Open AccessAspects of Tempo and Rhythmic Elaboration in Hindustani Music: A Corpus Study
This paper provides insights into aspects of tempo and rhythmic elaboration in Hindustani music, based on a study of a large corpus of recorded performances. Typical tempo developments and stress patterns within a metrical cycle are computed, which we refer to as tempo and rhythm patterns, respectively. Rhythm patterns are obtained by aggregating spectral features over metrical cycles. They reflect percussion patterns that are frequent in the corpus, and enable a discussion of the relation between such patterns and the underlying metrical framework, the taal. Tempo patterns, on the other hand, are computed using reference beat annotations. They document the dynamic development of tempo throughout a metrical cycle, and reveal insights into the flexibility of time in Hindustani music for the first time using quantitative methods on a large set of performances. Focusing on aspects of tempo and rhythm, we demonstrate the value of a computational methodology for the analysis of large music corpora by revealing the range of tempi used in performances, intra-cycle tempo dynamics and percussion accents at different positions of the taal cycle.
- 1 March 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 76-80
A generalized Bayesian model for tracking long metrical cycles in acoustic music signals
Most musical phenomena involve repetitive structures that enable listeners to track meter, i.e. the tactus or beat, the longer over-arching measure or bar, and possibly other related layers. Meters with long measure duration, sometimes lasting more than a minute, occur in many music cultures, e.g. from India, Turkey, and Korea. However, current meter tracking algorithms, which were devised for cycles of a few seconds length, cannot process such structures accurately. We present a novel generalization to an existing Bayesian model for meter tracking that overcomes this limitation. The proposed model is evaluated on a set of Indian Hindustani music recordings, and we document significant performance increase over the previous models. The presented model opens the way for computational analysis of performances with long metrical cycles, and has important applications in music studies as well as in commercial applications that involve such musics.
- 1 March 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 76-80
A generalized Bayesian model for tracking long metrical cycles in acoustic music signals
Most musical phenomena involve repetitive structures that enable listeners to track meter, i.e. the tactus or beat, the longer over-arching measure or bar, and possibly other related layers. Meters with long measure duration, sometimes lasting more than a minute, occur in many music cultures, e.g. from India, Turkey, and Korea. However, current meter tracking algorithms, which were devised for cycles of a few seconds length, cannot process such structures accurately. We present a novel generalization to an existing Bayesian model for meter tracking that overcomes this limitation. The proposed model is evaluated on a set of Indian Hindustani music recordings, and we document significant performance increase over the previous models. The presented model opens the way for computational analysis of performances with long metrical cycles, and has important applications in music studies as well as in commercial applications that involve such musics.
- 1 May 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15206149,p. 5217-5221
A supervised approach to hierarchical metrical cycle tracking from audio music recordings
A supervised approach to metrical cycle tracking from audio is presented, with a main focus on tracking the tala, the hierarchical cyclic metrical structure in Carnatic music. Given the tala of a piece, we aim to estimate the aksara (lowest metrical pulse), the aksara period, and the sama (first pulse of the tala cycle). Starting with percussion enhanced audio, we estimate the aksara pulse period from a tempogram computed using an onset detection function. A novelty function is computed using a self similarity matrix constructed using frame level audio features. These are then used to estimate possible aksara and sama candidates, followed by a candidate selection based on periodicity constraints, which leads to the final estimates. The algorithm is tested on an annotated collection of 176 pieces spanning four different talas. Though applied to Carnatic music, the framework presented is general and can be extended to other music cultures with cyclical metrical structures.
- 1 May 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15206149,p. 5217-5221
A supervised approach to hierarchical metrical cycle tracking from audio music recordings
A supervised approach to metrical cycle tracking from audio is presented, with a main focus on tracking the tala, the hierarchical cyclic metrical structure in Carnatic music. Given the tala of a piece, we aim to estimate the aksara (lowest metrical pulse), the aksara period, and the sama (first pulse of the tala cycle). Starting with percussion enhanced audio, we estimate the aksara pulse period from a tempogram computed using an onset detection function. A novelty function is computed using a self similarity matrix constructed using frame level audio features. These are then used to estimate possible aksara and sama candidates, followed by a candidate selection based on periodicity constraints, which leads to the final estimates. The algorithm is tested on an annotated collection of 176 pieces spanning four different talas. Though applied to Carnatic music, the framework presented is general and can be extended to other music cultures with cyclical metrical structures.
- Open Access
- 1 May 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 13 (15206149) , 2159-2163
A study of instrument-wise onset detection in Beijing Opera percussion ensembles
Note onset detection and instrument recognition are two of the most investigated tasks in Music Information Retrieval (MIR). Various detection methods have been proposed in previous research for western music, with less focus on other music cultures of the world. In this paper, we focus on onset detection for percussion instruments in Beijing Opera, a major genre of Chinese traditional music. A dataset of individual audio samples of four primary percussion instruments is used to obtain the spectral bases for each instrument. With these bases, we separate the input percussion ensemble recordings into its spectral sources and their activations using a Non-negative Matrix Factorization (NMF) based algorithm. A simple onset detection conducted on each NMF activation presents satisfactory overall detection rates, and provides us valuable implications and suggestions for future development of drum transcription and percussion pattern analysis in Beijing Opera.
- 1 May 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 13 (15206149) , 2159-2163
Open AccessA study of instrument-wise onset detection in Beijing Opera percussion ensembles
Note onset detection and instrument recognition are two of the most investigated tasks in Music Information Retrieval (MIR). Various detection methods have been proposed in previous research for western music, with less focus on other music cultures of the world. In this paper, we focus on onset detection for percussion instruments in Beijing Opera, a major genre of Chinese traditional music. A dataset of individual audio samples of four primary percussion instruments is used to obtain the spectral bases for each instrument. With these bases, we separate the input percussion ensemble recordings into its spectral sources and their activations using a Non-negative Matrix Factorization (NMF) based algorithm. A simple onset detection conducted on each NMF activation presents satisfactory overall detection rates, and provides us valuable implications and suggestions for future development of drum transcription and percussion pattern analysis in Beijing Opera.
- 2 January 2014
- Published by Taylor & Francis in Journal of New Music Research
- Vol. 43 (1) , 94-114
In Search of Automatic Rhythm Analysis Methods for Turkish and Indian Art Music
The aim of this paper is to identify and discuss various methods in computational rhythm description of Carnatic and Hindustani music of India, and Makam music of Turkey. We define and describe three relevant rhythm annotation tasks for these cultures—beat tracking, meter estimation, and downbeat detection. We then evaluate several methodologies from the state of the art in Music Information Retrieval (MIR) for these tasks, using manually annotated datasets of Turkish and Indian music. This evaluation provides insights into the nature of rhythm in these cultures and the challenges to automatic rhythm analysis. Our results indicate that the performance of evaluated approaches is not adequate for the presented tasks, and that methods that are suitable to tackle the culture specific challenges in computational analysis of rhythm need to be developed. The results from the different analysis methods enable us to identify promising directions for an appropriate exploration of rhythm analysis in Turkish, Carnatic and Hindustani music.
- 2 January 2014
- Published by Taylor & Francis in Journal of New Music Research
- Vol. 43 (1) , 94-114
In Search of Automatic Rhythm Analysis Methods for Turkish and Indian Art Music
The aim of this paper is to identify and discuss various methods in computational rhythm description of Carnatic and Hindustani music of India, and Makam music of Turkey. We define and describe three relevant rhythm annotation tasks for these cultures—beat tracking, meter estimation, and downbeat detection. We then evaluate several methodologies from the state of the art in Music Information Retrieval (MIR) for these tasks, using manually annotated datasets of Turkish and Indian music. This evaluation provides insights into the nature of rhythm in these cultures and the challenges to automatic rhythm analysis. Our results indicate that the performance of evaluated approaches is not adequate for the presented tasks, and that methods that are suitable to tackle the culture specific challenges in computational analysis of rhythm need to be developed. The results from the different analysis methods enable us to identify promising directions for an appropriate exploration of rhythm analysis in Turkish, Carnatic and Hindustani music.