Description:
Localizing a sound source is a fundamental but still challenging issue in many applications, where sound information is gathered by static and local microphone sensors. Therefore, this work proposes a new system by exploiting advances in sensor networks and robotics to more accurately address the problem of sound source localization. By the use of the network infrastructure, acoustic sensors are more efficient to spatially monitor acoustical phenomena. Furthermore, a mobile robot is proposed to carry an extra microphone array in order to collect more acoustic signals when it travels around the environment. Driving the robot is guided by the need to increase the quality of the data gathered by the static acoustic sensors, which leads to better probabilistic fusion of all the information gained, so that an increasingly accurate map of the sound source can be built. The proposed system has been validated in a real-life environment, where the obtained results are highly promising.
Description:
Six male music theatre singers were recorded in three different voice qualities: legit and two types of belt ("chesty" and "twangy"), on two vowels ([e] and [O]), at four increasing pitches in the upper limit of each singer's belt range (similar to 250-440 Hz). The audio signal, the electroglottographic (EGG) signal, and the vocal tract impedance were all measured simultaneously. Voice samples were analyzed and then evaluated perceptually by 16 expert listeners. The three qualities were produced with significant differences at the physiological, acoustical, and perceptual levels: Singers produced belt qualities with a higher EGG contact quotient (CQ(EGG)) and greater contacting speed quotient (Qcs), greater sound pressure level (SPL), and energy above 1 kHz (alpha ratio), and with higher frequencies of the first two vocal tract resonances (f(R1), f(R2)), especially in the upper pitch range when compared to legit. Singers produced the chesty belt quality with higher CQ(EGG), Qcs, and SPL values and lower alpha ratios over the whole belt range, and with higher f(R1) at the higher pitch range when compared to twangy belt. Consistent tuning of f(R1) to the second voice harmonic (2f(0)) was observed in all three qualities and for both vowels. Expert listeners tended to identify all qualities based on the same acoustical and physiological variations as those observed in the singers' intended qualities. (C) 2016 Acoustical Society of America.
Description:
Six male music theatre singers were recorded in three different voice qualities: legit and two types of belt ("chesty" and "twangy"), on two vowels ([e] and [O]), at four increasing pitches in the upper limit of each singer's belt range (similar to 250-440 Hz). The audio signal, the electroglottographic (EGG) signal, and the vocal tract impedance were all measured simultaneously. Voice samples were analyzed and then evaluated perceptually by 16 expert listeners. The three qualities were produced with significant differences at the physiological, acoustical, and perceptual levels: Singers produced belt qualities with a higher EGG contact quotient (CQ(EGG)) and greater contacting speed quotient (Qcs), greater sound pressure level (SPL), and energy above 1 kHz (alpha ratio), and with higher frequencies of the first two vocal tract resonances (f(R1), f(R2)), especially in the upper pitch range when compared to legit. Singers produced the chesty belt quality with higher CQ(EGG), Qcs, and SPL values and lower alpha ratios over the whole belt range, and with higher f(R1) at the higher pitch range when compared to twangy belt. Consistent tuning of f(R1) to the second voice harmonic (2f(0)) was observed in all three qualities and for both vowels. Expert listeners tended to identify all qualities based on the same acoustical and physiological variations as those observed in the singers' intended qualities. (C) 2016 Acoustical Society of America.
Description:
Blind Source Separation techniques are widely used in the field of wireless communication for a very long time to extract signals of interest from a set of multiple signals without training data. In this paper, we investigate the problem of separation of the human voice from a mixture of human voice and sounds from different musical instruments. The human voice may be a singing voice in a song or may be a part of some news, broadcast by a channel with background music. This paper proposes a generalized Short Time Fourier Transform (STFT)-based technique, combined with filter bank to extract vocals from background music. The main purpose is to design a filter bank and to eliminate background aliasing errors with best reconstruction conditions, having approximated scaling factors. Stereo signals in time-frequency domain are used in experiments. The input stereo signals are processed in the form of frames and passed through the proposed STFT-based technique. The output of the STFT-based technique is passed through the filter bank to minimize the background aliasing errors. For reconstruction, first an inverse STFT is applied and then the signals are reconstructed by the OverLap-Add method to get the final output, containing vocals only. The experiments show that the proposed approach performs better than the other state-of-the-art approaches, in terms of Signal-to-Interference Ratio (SIR) and Signal-to-Distortion Ratio (SDR), respectively.