Kiwan Eom

An integrated music recommendation system

In this paper, an integrated music recommendation system is proposed, which contains the functions of automatic music genre classification, automatic music emotion classification, and music similarity query. A novel tempo feature, named as log-scale modulation frequency coefficients, is presented in this paper. With AdaBoost algorithm, the proposed tempo feature is combined with timbre features and improves the performance of music genre and emotion classification. Comparing with the conventional methods based on timbre features, the precision of five-genre classification is enhanced from 86.8% to 92.2% and the accuracy of four-emotion classification is increased from 86.0% to 90.5%. Based on the results of music genre/emotion classification, we design a similarity query scheme, which can speed up the similarity query process without decreasing the precision. Furthermore, all the features employed in this paper are extracted from the data of MP3 partially decoding, which significantly reduces the feature extraction time

A Tempo Feature via Modulation Spectrum Analysis and its Application to Music Emotion Classification

This paper proposes a tempo feature extraction method based on the long-term modulation spectrum analysis. To transform the modulation spectrum to a condensed feature vector, the log-scale modulation frequency coefficients are introduced. This idea aims at averaging the modulation frequency energy via the constant-Q filter-banks. Further it is pointed out that the feature can be extracted directly from the perceptually compressed data of digital music archives. To verify the effectiveness of the feature and its utility to music applications, the feature vector is used in a music emotion classification system. The system consisting two layers of Adaboost classifiers. In the first layer the conventional timbre features are employed. Then by adding the tempo feature in the second layer, the classification precision is improved dramatically. By this way the discriminability of the classifier based on the given features can be exploited extremely. The system obtains high classification precision on a small corpus. It proves that the proposed feature is very effective and computationally efficient to characterize the tempo information of music

Switching linear dynamic transducer for stereo data based speech feature mapping

The performance of a speech recognition system may be degraded even without any background noise because of the linear or non-linear distortions incurred by recording devices or reverberations. One of the well-known approaches to reduce this channel distortion is feature mapping which maps the distorted speech feature to its clean counterpart. The feature mapping rule is usually trained based on a set of stereo data which consists of the simultaneous recordings obtained in both the reference and target conditions. In this paper, we propose a novel approach to speech feature sequence mapping based on the switching linear dynamic transducer (SLDT). The proposed algorithm enables us a sequence-to-sequence mapping in a systematic way, instead of the traditional vector-to-vector mapping. The proposed approach is applied to compensate channel distortion in speech recognition and shows improvement in recognition performance.

DOA Estimation for Wideband Signal: Multiple Frequency Bins Versus Multiple Sensors

In conventional direction of arrival (DOA) estimation techniques, the inter-sensor spacing of an array is designed according to the minimal wavelength of an applied signal to avoid spatial ambiguity. This constraint greatly restricts the resolution of array or the performance of DOA estimation for a wide band signal. In this work, a new concept of wideband DOA estimation is described. Compared with multiple signal classification (MUSIC) method, the proposed method in this work uses multiple frequency bins instead of multiple sensors to form the steering vector, which reduces the sensor number for DOA estimation into 2. Besides, in this wideband DOA estimation method, the inter-sensor spacing is determined by the frequency resolution rather than the highest frequency (or minimal wavelength), so this method greatly decreases the restriction on sensor interval. Simulation results show that the proposed method with only two sensors can achieve comparative performance with convtmtional DOA estimation method using more than 20 sensors.

Data-driven lexicon refinement using local and web resources for Chinese speech recognition

This paper proposes a data-driven lexicon refinement method. By expanding and polishing lexicon using local and web resources, accuracy of Chinese automatic speech recognition (ASR) system is boosted effectively. The proposed lexicon refining process is composed of two steps. First, an improved intra-word measure is introduced. It helps to expand lexicon from local text corpora. Second, the expanded lexicon is polished by enumerating the popularity of appended words based on web query results via search engine. The evaluation experiments are carried out on an application of voice-enabled tourist information query system. Experimental results show that the proposed lexicon refinement method reduces character error rate (CER) by 7.9% relatively.