Turkish, being an agglutinative language with rich morphology, presents challenges for Large Vocabulary Continuous Speech Recognition (LVCSR) systems. First, the agglutinative nature of Turkish leads to a high number of Out-of Vocabulary (OOV) words which in turn lower Automatic Speech Recognition (ASR) accuracy. Second, Turkish has a relatively free word order that leads to non-robust language model estimates. These challenges have been mostly handled by using meaningful segmentations of words, called sub-lexical units, in language modeling. However, a shortcoming of sub-lexical units is over-generation which needs to be dealt with for higher accuracies. This dissertation aims to address the challenges of Turkish in LVCSR. Grammatical and statistical sub-lexical units for language modeling are investigated and they yield substantial improvements over the word language models. Our novel approach inspired by dynamic vocabulary adaptation mostly recovers the errors caused by over-generation and ...

The thesis deals with automatic interpretation of naturally spoken utterances for limited-domain applications. Specifically, the problem is examined by means of a dialogue system for an airport information application. In contrast to traditional two-stage systems, speech recognition and semantic processing are tightly coupled. This avoids interpretation errors due to early decisions. The presented one-stage decoding approach utilizes a uniform, stochastic knowledge representation based on weighted transition network hierarchies, which describe phonemes, words, word classes and semantic concepts. A robust semantic model, which is estimated by combination of data-driven and rule-based approaches, is part of this representation. The investigation of this hierarchical language model is the focus of this work. Furthermore, methods for modeling out-of-vocabulary words and for evaluating semantic trees are introduced.

One desideratum in designing cognitive robots is autonomous learning of communication skills, just like humans. The primary step towards this goal is vocabulary acquisition. Being different from the training procedures of the state-of-the-art automatic speech recognition (ASR) systems, vocabulary acquisition cannot rely on prior knowledge of language in the same way. Like what infants do, the acquisition process should be data-driven with multi-level abstraction and coupled with multi-modal inputs. To avoid lengthy training efforts in a word-by-word interactive learning process, a clever learning agent should be able to acquire vocabularies from continuous speech automatically. The work presented in this thesis is entitled \emph{Constrained Non-negative Matrix Factorization for Vocabulary Acquisition from Continuous Speech}. Enlightened by the extensively studied techniques in ASR, we design computational models to discover and represent vocabularies from continuous speech with little prior knowledge of the language to ...

When exposed to noise, speakers will modify the way they speak in an effort to maintain intelligible communication. This process, which is referred to as Lombard effect (LE), involves a combination of both conscious and subconscious articulatory adjustment. Speech production variations due to LE can cause considerable degradation in automatic speech recognition (ASR) since they introduce a mismatch between parameters of the speech to be recognized and the ASR system’s acoustic models, which are usually trained on neutral speech. The main objective of this thesis is to analyze the impact of LE on speech production and to propose methods that increase ASR system performance in LE. All presented experiments were conducted on the Czech spoken language, yet, the proposed concepts are assumed applicable to other languages. The first part of the thesis focuses on the design and acquisition of a ...

This thesis presents a system for the interpretation of natural speech which serves as input module for a spoken dialog system. It carries out the task of extracting application-specific pieces of information from the user utterance in order to pass them to the control module of the dialog system. By following the approach of integrating speech recognition and speech interpretation, the system is able to determine the spoken word sequence together with the hierarchical utterance structure that is necessary for the extraction of information directly from the recorded speech signal. The efficient implementation of the underlying decoder is based on the powerful tool of weighted finite state transducers (WFSTs). This tool allows to compile all involved knowledge sources into an optimized network representation of the search space which is constructed dynamically during the ongoing decoding process. In addition to the ...

Con dence measures for the results of speech/speaker recognition make the systems more useful in the real time applications. Con dence measures provide a test statistic for accepting or rejecting the recognition hypothesis of the speech/speaker recognition system. Speech/speaker recognition systems are usually based on statistical modeling techniques. In this thesis we de ned con dence measures for statistical modeling techniques used in speech/speaker recognition systems. For speech recognition we tested available con dence measures and the newly de ned acoustic prior information based con dence measure in two di erent conditions which cause errors: the out-of-vocabulary words and presence of additive noise. We showed that the newly de ned con dence measure performs better in both tests. Review of speech recognition and speaker recognition techniques and some related statistical methods is given through the thesis. We de ned also ...

Speech is at the core of human communication. Speaking and listing comes so natural to us that we do not have to think about it at all. The underlying cognitive processes are very rapid and almost completely subconscious. It is hard, if not impossible not to understand speech. For computers on the other hand, recognising speech is a daunting task. It has to deal with a large number of different voices "influenced, among other things, by emotion, moods and fatigue" the acoustic properties of different environments, dialects, a huge vocabulary and an unlimited creativity of speakers to combine words and to break the rules of grammar. Almost all existing automatic speech recognisers use statistics over speech sounds "what is the probability that a piece of audio is an a-sound" and statistics over word combinations to deal with this complexity. The ...

This thesis investigates the application of language models based on semantic similarity to Automatic Speech Recognition for meetings. We consider data-driven Latent Semantic Analysis based and knowledge-driven WordNet-based models. Latent Semantic Analysis based models are trained for several background domains and it is shown that all background models reduce perplexity compared to the n-gram baseline models, and some background models also significantly improve speech recognition for meetings. A new method for interpolating multiple models is introduced and the relation to cache-based models is investigated. The semantics of the models is investigated through a synonymity task. WordNet-based models are defined for different word-word similarities that use information encoded in the WordNet graph and corpus information. It is shown that these models can significantly improve over baseline random models on the task of word prediction, and that the chosen part-of-speech context is ...

Despite the outstanding progress made on automatic speech recognition (ASR) throughout the last decades, noise-robust ASR still poses a challenge. Tackling with acoustic noise in ASR systems is more important than ever before for a twofold reason: 1) ASR technology has begun to be extensively integrated in intelligent mobile devices (IMDs) such as smartphones to easily accomplish different tasks (e.g. search-by-voice), and 2) IMDs can be used anywhere at any time, that is, under many different acoustic (noisy) conditions. On the other hand, with the aim of enhancing noisy speech, IMDs have begun to embed small microphone arrays, i.e. microphone arrays comprised of a few sensors close each other. These multi-sensor IMDs often embed one microphone (usually at their rear) intended to capture the acoustic environment more than the speaker’s voice. This is the so-called secondary microphone. While classical microphone ...

This thesis addresses the problem of vision based sign language recognition and focuses on three main tasks to design improved techniques that increase the performance of sign language recognition systems. We first attack the markerless tracking problem during natural and unrestricted signing in less restricted environments. We propose a joint particle filter approach for tracking multiple identical objects, in our case the two hands and the face, which is robust to situations including fast movement, interactions and occlusions. Our experiments show that the proposed approach has a robust tracking performance during the challenging situations and is suitable for tracking long durations of signing with its ability of fast recovery. Second, we attack the problem of the recognition of signs that include both manual (hand gestures) and non-manual (head/body gestures) components. We investigated multi-modal fusion techniques to model the different temporal ...

In natural listening environments, speech signals are easily distorted by variousacoustic interference, which reduces the speech quality and intelligibility of human listening; meanwhile, it makes difficult for many speech-related applications, such as automatic speech recognition (ASR). Thus, many speech enhancement (SE) algorithms have been developed in the past decades. However, most current SE algorithms are difficult to capture underlying speech information (e.g., phoneme) in the SE process. This causes it to be challenging to know what specific information is lost or interfered with in the SE process, which limits the application of enhanced speech. For instance, some SE algorithms aimed to improve human listening usually damage the ASR system. The objective of this dissertation is to develop SE algorithms that have the potential to capture various underlying speech representations (information) and improve the quality and intelligibility of noisy speech. This ...

In the audio research field, the majority of machine learning systems focus on recognizing a limited number of sound events. However, when a machine interacts with real data, it must be able to handle much more varied and complex situations. To tackle this problem, annotators use natural language, which allows any sound information to be summarized. Automated Audio Captioning (AAC) was introduced recently to develop systems capable of automatically producing a description of any type of sound in text form. This task concerns all kinds of sound events such as environmental, urban, domestic sounds, sound effects, music or speech. This type of system could be used by people who are deaf or hard of hearing, and could improve the indexing of large audio databases. In the first part of this thesis, we present the state of the art of the ...

In a natural acoustic environment, multiple sources are usually active at the same time. The task of source separation is the estimation of individual source signals from this complex mixture. The challenge of single channel source separation (SCSS) is to recover more than one source from a single observation. Basically, SCSS can be divided in methods that try to mimic the human auditory system and model-based methods, which find a probabilistic representation of the individual sources and employ this prior knowledge for inference. This thesis presents several strategies for the separation of two speech utterances mixed into a single channel and is structured in four parts: The first part reviews factorial models in model-based SCSS and introduces the soft-binary mask for signal reconstruction. This mask shows improved performance compared to the soft and the binary masks in automatic speech recognition ...

Much like natural language, music is highly structured, with strong priors on the likelihood of note sequences. In automatic speech recognition (ASR), these priors are called language models, which are used in addition to acoustic models and participate greatly to the success of today's systems. However, in Automatic Music Transcription (AMT), ASR's musical equivalent, Music Language Models (MLMs) are rarely used. AMT can be defined as the process of extracting a symbolic representation from an audio signal, describing which notes were played at what time. In this thesis, we investigate the design of MLMs using recurrent neural networks (RNNs) and their use for AMT. We first look into MLM performance on a polyphonic prediction task. We observe that using musically-relevant timesteps results in desirable MLM behaviour, which is not reflected in usual evaluation metrics. We compare our model against benchmark ...

Recent advances in computer technology and speech and language processing have made possible that some new ways of person-machine communication and computer assistance to human activities start to appear feasible. Concretely, the interest on the development of new challenging applications in indoor environments equipped with multiple multimodal sensors, also known as smart-rooms, has considerably grown. In general, it is well-known that the quality of speech signals captured by microphones that can be located several meters away from the speakers is severely distorted by acoustic noise and room reverberation. In the context of the development of hands-free speech applications in smart-room environments, the use of obtrusive sensors like close-talking microphones is usually not allowed, and consequently, speech technologies must operate on the basis of distant-talking recordings. In such conditions, speech technologies that usually perform reasonably well in free of noise and ...