• Speech Sciences
• /
• v.8 no.2
• /
• pp.83-90
• /
• 2001

We proposed a probabilistic approach of using syllable bigrams to the word-spacing problem. Syllable bigrams are extracted and the frequencies are calculated for the large corpus of 12 million words. Based on the syllable bigrams, we performed three experiments: (1) automatic word-spacing, (2) detection and correction of word-spacing errors for spelling checker, and (3) automatic insertion of a space at the end of line in the character recognition system. Experimental results show that the accuracy ratios are 97.7 percent, 82.1 percent, and 90.5%, respectively.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.1
• /
• pp.51-55
• /
• 2019

This paper proposes a speech enhancement method utilizing the feature compensation gain for robust speech recognition performances in noisy environments. In this paper we propose a speech enhancement method utilizing the feature compensation gain which is obtained from the PCGMM (Parallel Combined Gaussian Mixture Model)-based feature compensation method employing variational model composition. The experimental results show that the proposed method significantly outperforms the conventional front-end algorithms and our previous research over various background noise types and SNR (Signal to Noise Ratio) conditions in mismatched ASR (Automatic Speech Recognition) system condition. The computation complexity is significantly reduced by employing the noise model selection technique with maintaining the speech recognition performance at a similar level.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.8
• /
• pp.105-113
• /
• 1996

In this paper, a new korean digita speech recognition technique was proposed using muktolayer perceptron (MLP). In spite of its weakness in dynamic signal recognition, MLP was adapted for this model, cecause korean syllable could give static features. It is so simle in its structure and fast in its computing that MLP was used to the suggested system. MLP's input vectors was transformed using karhunen-loeve transformation (KLT), which compress signal successfully without losin gits separateness, but its physical properties is changed. Because the suggested technique could extract static features while it is not affected from the changes of syllable lengths, it is effectively useful for korean numeric recognition system. Without decreasing classification rates, we can save the time and memory size for computation using KLT. The proposed feature extraction technique extracts same size of features form the tow same parts, front and end of a syllable. This technique makes frames, where features are extracted, using unique size of windows. It could be applied for continuous speech recognition that was not easy for the normal neural network recognition system.

• Proceedings of the KSPS conference
• /
• 2004.05a
• /
• pp.231-234
• /
• 2004

This paper describes an efficient method for improving the noise-robustness in speech recognition in a running car by considering wind noise. In driving car, mainly three kind of noises engine noise, tire noise and wind noise, are severely affect recognition performance. Especially wind noise is an important factor in driving car with window opened. We analyzed wind noise in various driving conditions that are 60, 80, 100 km/h with window fully opened, window half opened. We clarified that the recognition rate is significantly degenerated when the wind noise components in the frequency range above 200 Hz are large. We developed a preprocessing method to improve the noise robustness despite of wind noise. We adaptively changed the cutoff frequency of the front-end high-pass filter from 100 through 200 Hz according to the level of the wind noise components. By this method, the recognition rate is considerably improved for all kind of driving conditions

• International Journal of Contents
• /
• v.15 no.4
• /
• pp.59-64
• /
• 2019

In recent years, emotion recognition has been an interesting and challenging topic. Compared to facial expressions and speech modality, gesture-based emotion recognition has not received much attention with only a few efforts using traditional hand-crafted methods. These approaches require major computational costs and do not offer many opportunities for improvement as most of the science community is conducting their research based on the deep learning technique. In this paper, we propose an end-to-end deep learning approach for classifying emotions based on bodily gestures. In particular, the informative keyframes are first extracted from raw videos as input for the 3D-CNN deep network. The 3D-CNN exploits the short-term spatiotemporal information of gesture features from selected keyframes, and the convolutional LSTM networks learn the long-term feature from the features results of 3D-CNN. The experimental results on the FABO dataset exceed most of the traditional methods results and achieve state-of-the-art results for the deep learning-based technique for gesture-based emotion recognition.

• Speech Sciences
• /
• v.10 no.1
• /
• pp.95-105
• /
• 2003

This paper describes a new parameter for voice activity detection which serves as a front-end part for automatic speech recognition systems. The new parameter called run-ratio is derived from the runs test statistic which is used in the statistical test for randomness of a given sequence. The run-ratio parameter has the property that the values of the parameter for the random sequence are about 1. To apply the run-ratio parameter into the voice activity detection method, it is assumed that the samples of an inputted audio signal should be converted to binary sequences of positive and negative values. Then, the silence region in the audio signal can be regarded as random sequences so that their values of the run-ratio would be about 1. The run-ratio for the voiced region has far lower values than 1 and for fricative sounds higher values than 1. Therefore, the parameter can discriminate speech signals from the background sounds by using the newly derived run-ratio parameter. The proposed voice activity detector outperformed the conventional energy-based detector in the sense of error mean and variance, small deviation from true speech boundaries, and low chance of missing real utterances

• Journal of Convergence for Information Technology
• /
• v.10 no.8
• /
• pp.35-39
• /
• 2020

DNN error is small compared to the conventional speech recognition system, DNN is difficult to parallel training, often the amount of calculations, and requires a large amount of data obtained. In this paper, we generate a phoneme unit to estimate the GMM parameters with each phoneme model parameters from the GMM to solve the problem efficiently. And it suggests ways to improve performance through clustering for a specific vocabulary to effectively apply them. To this end, using three types of word speech database was to have a DB build vocabulary model, the noise processing to extract feature with Warner filters were used in the speech recognition experiments. Results using the proposed method showed a 97.9% recognition rate in speech recognition. In this paper, additional studies are needed to improve the problems of improved over fitting.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.1
• /
• pp.53-60
• /
• 1999

This paper is a study on the korean broadcasting speech recognition. Here we present the methods for the large vocabuary continuous speech recognition. Our main concerns are the language modeling and the search algorithm. The used acoustic model is the uni-phone semi-continuous hidden markov model and the used linguistic model is the N-gram model. The search algorithm consist of three phases in order to utilize all available acoustic and linguistic information. First, we use the forward Viterbi beam search to find word end frames and to estimate related scores. Second, we use the backword Viterbi beam search to find word begin frames and to estimate related scores. Finally, we use A/sup */ search to combine the above two results with the N-grams language model and to get recognition results. Using these methods maximum 96.0% word recognition rate and 99.2% syllable recognition rate are achieved for the speaker-independent continuous speech recognition problem with about 12,000 vocabulary size.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.6
• /
• pp.284-290
• /
• 2024

Speech emotion recognition (SER) is a technique that is used to analyze the speaker's voice patterns, including vibration, intensity, and tone, to determine their emotional state. There has been an increase in interest in artificial intelligence (AI) techniques, which are now widely used in medicine, education, industry, and the military. Nevertheless, existing researchers have attained impressive results by utilizing acted-out speech from skilled actors in a controlled environment for various scenarios. In particular, there is a mismatch between acted and spontaneous speech since acted speech includes more explicit emotional expressions than spontaneous speech. For this reason, spontaneous speech-emotion recognition remains a challenging task. This paper aims to conduct emotion recognition and improve performance using spontaneous speech data. To this end, we implement deep learning-based speech emotion recognition using the VGG (Visual Geometry Group) after converting 1-dimensional audio signals into a 2-dimensional spectrogram image. The experimental evaluations are performed on the Korean spontaneous emotional speech database from AI-Hub, consisting of 7 emotions, i.e., joy, love, anger, fear, sadness, surprise, and neutral. As a result, we achieved an average accuracy of 83.5% and 73.0% for adults and young people using a time-frequency 2-dimension spectrogram, respectively. In conclusion, our findings demonstrated that the suggested framework outperformed current state-of-the-art techniques for spontaneous speech and showed a promising performance despite the difficulty in quantifying spontaneous speech emotional expression.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.439-451
• /
• 2021

We propose a deep learning-based beamformer combined with spectral subtraction for continuous speech recognition operating in noisy environments. Conventional beamforming systems were mostly evaluated by using pre-segmented audio signals which were typically generated by mixing speech and noise continuously on a computer. However, since speech utterances are sparsely uttered along the time axis in real environments, conventional beamforming systems degrade in case when noise-only signals without speech are input. To alleviate this drawback, we combine online beamforming algorithm and spectral subtraction. We construct a Continuous Speech Enhancement (CSE) evaluation set to evaluate the online beamforming algorithm in noisy environments. The evaluation set is built by mixing sparsely-occurring speech utterances of the CHiME3 evaluation set and continuously-played CHiME3 background noise and background music of MUSDB. Using a Kaldi-based toolkit and Google web speech recognizer as a speech recognition back-end, we confirm that the proposed online beamforming algorithm with spectral subtraction shows better performance than the baseline online algorithm.