Musical Instrument Sounds Research Articles

Identification of a dominating instrument in polytimbral same-pitch mixes using SVM classifiers with non-linear kernel

In this paper we deal with the problem of identification of the dominating musical instrument in a recording containing simultaneous sounds of the same pitch. Sustained harmonic sounds from one octave of twelve instruments were considered. The training data set contains isolated sounds of two forms, one from selected musical instruments, and the other from the same mixed with artificial harmonic and noise sounds of lower amplitude. The test data set contains mixes of musical instrument sounds. A Support Vector Machine classifier was used for training and testing experiments, using a non-linear kernel. Additionally, we performed tests on data based on different recordings of instruments than those used in the training procedure described above. Results of these experiments are presented and discussed.

Enhancing method for speech mixed with musical instrument sounds based on sinusoidal trajectories

Enhancing method for speech mixed with musical instrument sounds based on sinusoidal trajectories

Detection of time-varying harmonic amplitude alterations due to spectral interpolations between musical instrument tones

Gradated spectral interpolations between musical instrument tone pairs were used to investigate discrimination as a function of time-averaged spectral difference. All possible nonidentical pairs taken from a collection of eight musical instrument sounds consisting of bassoon, clarinet, flute, horn, oboe, saxophone, trumpet, and violin were tested. For each pair, several tones were generated with different balances between the primary and secondary instruments, where the balance was fixed across the duration of each tone. Among primary instruments it was found that changes to horn and bassoon [corrected] were most easily discriminable, while changes to saxophone and trumpet timbres were least discriminable. Among secondary instruments, the clarinet had the strongest effect on discrimination, whereas the bassoon had the least effect. For primary instruments, strong negative correlations were found between discrimination and their spectral incoherences, suggesting that the presence of dynamic spectral variations tends to increase the difficulty of detecting time-varying alterations such as spectral interpolation.

Nondestructive characterization of musical pillars of Mahamandapam of Vitthala Temple at Hampi, India

This paper presents the first scientific investigation on the musical pillars of the Vitthala Temple at Hampi, India. The solid stone columns in these pillars produce audible sound, when struck with a finger. Systematic investigations on the acoustic characteristics of the musical pillars of mahamandapam (great stage) of the Vitthala Temple have been carried out. The 11 most popular pillars that produce sounds of specific musical instruments are considered for the investigations. The sound produced from these 11 most popular musical pillars was recorded systematically and different nondestructive testing techniques such as low frequency ultrasonic testing, impact echo testing, and in situ metallography were employed on the musical columns of these pillars. The peak frequencies in the amplitude spectrum of the sound produced from various columns in these pillars are correlated with the dimensional measurements and ultrasonic velocity determined using impact echo technique. The peak frequencies obtained experimentally have been found to have excellent correlation with the calculated flexural frequencies based on the dimensional measurements and ultrasonic velocities of the columns.

Live computer‐based demonstrations in musical acoustics education

Live demonstrations play a vital pedagogical role in any science education programme. For a subject as interdisciplinary as musical acoustics, embracing as it does mathematics, physics, engineering, psychology, cognitive science, neurophysiology, musicology and music technology, demonstrations of psychoacoustical phenomena and performances on musical instruments provide invaluable methods for involving and intriguing a general audience. The recent dramatic increase in the memory and speed of laptop computers has opened new possibilities for live computer‐based demonstrations. This talk explores some of these possibilities, including demonstrations of the analysis and display of musical instrument sounds using real‐time spectrographic programs and the explanation of important musical instrument characteristics using physical modelling synthesis.

Spectral envelope sensitivity of musical instrument sounds

It is well known that the spectral envelope is a perceptually salient attribute in musical instrument timbre perception. While a number of studies have explored discrimination thresholds for changes to the spectral envelope, the question of how sensitivity varies as a function of center frequency and bandwidth for musical instruments has yet to be addressed. In this paper a two-alternative forced-choice experiment was conducted to observe perceptual sensitivity to modifications made on trumpet, clarinet and viola sounds. The experiment involved attenuating 14 frequency bands for each instrument in order to determine discrimination thresholds as a function of center frequency and bandwidth. The results indicate that perceptual sensitivity is governed by the first few harmonics and sensitivity does not improve when extending the bandwidth any higher. However, sensitivity was found to decrease if changes were made only to the higher frequencies and continued to decrease as the distorted bandwidth was widened. The results are analyzed and discussed with respect to two other spectral envelope discrimination studies in the literature as well as what is predicted from a psychoacoustic model.

Toward Next-Generation Digital Keyboard Instruments

In this article, alternative approaches to digital keyboard instrument synthesis are looked into. Physics-based sound synthesis, which aims at generating natural-sounding musical instrument tones algorithmically without using a large sample database, is a promising approach. It would provide high-quality music synthesis to systems that cannot afford a large memory, such as mobile phones and portable electronic games. The realistic parametric synthesis of musical instrument sounds is still a challenge, but physical modeling techniques introduced during the last few decades can help to solve it. Recently, the first commercial products have been introduced, for example, by Pianoteq. Three keyboard instruments, the clavichord, the harpsichord, and the grand piano, are focused on here. The sound production principles and acoustics of these instruments are first discussed. Then, the previous parametric synthesis algorithms developed for these instruments are reviewed. The remaining part of this article concentrates on new signal processing methods for parametric synthesis of the piano

Cerebral response to ‘voiceness’: a functional magnetic resonance imaging study

We evaluated the response of the voice-selective areas of the auditory cortex to sound 'voiceness', that is, the degree to which an auditory stimulus resembles human voice. Normal participants were scanned using event-related functional magnetic resonance imaging while passively listening to stimuli drawn from a 'voiceness' continuum generated via auditory morphing between sounds of voice and sounds of musical instruments. The voice-selective areas of the left and right superior temporal sulcus did not show the expected relation between 'voiceness' and size effect. Instead, superior temporal sulcus activity seemed mostly driven by sound naturalness, with largest activity differences observed for the intermediate, voice-instrument hybrid stimuli.

Multidimensional scaling analysis of centroid- and attack/decay-normalized musical instrument sounds

Ten sustained musical instrument tones (bassoon, cello, clarinet, flute, horn, oboe, recorder, alto saxophone, trumpet, and violin) were spectrally analyzed and then equalized for duration, attack and decay time, fundamental frequency, number of harmonics, average spectral centroid, and presentation loudness. The tones were resynthesized both with time-varying harmonic amplitudes and frequencies (dynamic case) and fixed amplitudes and frequencies (static case). Tone triads were presented to ten musically experienced listeners whose tasks were to specify the most dissimilar and most similar pairs in each triad. Based on the resulting dissimilarity matrix, multidimensional scaling (MDS) was used to position the instruments in two- and three-dimensional metric spaces. Two measures of instrument amplitude spectra were found to correlate strongly with MDS dimensions. For both the static- and dynamic-case 2-D solutions, the ratio of even-to-odd rms amplitudes correlated strongly with one of the dimensions. For the dynamic case, spectral centroid variation correlated strongly with the second dimension. Also, 2-D solution instrument groupings agreed well with groupings based on coefficients of the first two components of a principle components analysis representing 90% of the instruments’ spectral variance. [This work was supported by the Hong Kong Research Grants Council’s CERG Project 613505.]

Introduction of the Real World Computing music database

This paper introduces the RWC (Real World Computing) Music Database, a copyright-cleared music database that is available to researchers as a common foundation for research. Shared databases are common in other research fields and have contributed importantly to progress in those fields. The field of music information processing, however, has lacked a common database of musical pieces and a large-scale database of musical instrument sounds. The RWC Music Database was therefore built in fiscal 2000 and 2001 as the world’s first large-scale music database compiled specifically for research purposes. It contains six original collections: the Popular Music Database (100 pieces), Royalty-Free Music Database (15 pieces), Classical Music Database (50 pieces), Jazz Music Database (50 pieces), Music Genre Database (100 pieces), and Musical Instrument Sound Database (50 instruments). To address copyright issues, all 315 musical pieces were originally composed, arranged, or performed, and all instrumental sounds were originally recorded. The database has already been distributed to more than 200 research groups and is widely used. In addition, a continuous effort has been undertaken to manually annotate a set of music-scene descriptions for the musical pieces, called AIST Annotation, which consists of the beat structure, melody line, and chorus sections.

Timbre representation for automatic classification of musical instruments

Human communication includes the capability of recognition. This is particularly true of auditory communication. Music information retrieval (MIR) turns out to be particularly challenging, since many problems remain still unsolved. Topics that should be included within the scope of MIR are automatic classification of musical instruments/phrases/styles, music representation and indexing, estimating musical similarity using both perceptual and musicological criteria, recognizing music using audio and/or semantic description, language modeling for music, auditory scene analysis, and others. Many features of music content description are based on perceptual phenomena and cognition. However, it can easily be observed that most of the low-level descriptors used, for example, in musical instrument classification are more data- than human-oriented. This is because the idea behind these features is to have data defined and linked in such a way as to be able to use it for more effective automatic discovery, integration, and reuse in various applications. The ambitious task is, however, to provide seamless meaning to low- and high-level descriptors such as timbre descriptors and linking them together. In such a way data can be processed and shared by both systems and people. This paper presents a study related to timbre representation of musical instrument sounds.

Perception of acoustic scale and size in musical instrument sounds.

There is size information in natural sounds. For example, as humans grow in height, their vocal tracts increase in length, producing a predictable decrease in the formant frequencies of speech sounds. Recent studies have shown that listeners can make fine discriminations about which of two speakers has the longer vocal tract, supporting the view that the auditory system discriminates changes on the acoustic-scale dimension. Listeners can also recognize vowels scaled well beyond the range of vocal tracts normally experienced, indicating that perception is robust to changes in acoustic scale. This paper reports two perceptual experiments designed to extend research on acoustic scale and size perception to the domain of musical sounds: The first study shows that listeners can discriminate the scale of musical instrument sounds reliably, although not quite as well as for voices. The second experiment shows that listeners can recognize the family of an instrument sound which has been modified in pitch and scale beyond the range of normal experience. We conclude that processing of acoustic scale in music perception is very similar to processing of acoustic scale in speech perception.

Amplitude fluctuations within musical instrument sounds because of impulse reflections

Musical instrument sounds nearly always show amplitude fluctuations in their overtones within the quasisteady state of the sound.These fluctuations make the sound more lively and are found in blown instruments like the saxophone or clarinet, in string instruments like the guitar or the violin, or in percussion instruments. The reason for this can be found within a feature common to all musical instruments, their impulse-driven behavior. Not only do the blown instruments work with impulses; the string forces of plucked, strucked, or bowed instruments of the top plate are also an impulse train. So, when looking at the system during the steady state not like one of standing waves but of impulses being reflected at different inhomogeneties within the structure, these reflections can be shown to be the reason for the amplitude fluctuations. The mechanism is caused by the fact that the energy within an impulse is split into potential and kinetic energy, where only the kinetic energy is radiated. When traveling along a homogeneous region, the energies are more or less conserved. But, during reflections, these energies get larger and smaller and cause a fluctuation in the kinetic energy and so in the radiation of sound.

Pitch-Dependent Identification of Musical Instrument Sounds

This paper describes a musical instrument identification method that takes into consideration the pitch dependency of timbres of musical instruments. The difficulty in musical instrument identification resides in the pitch dependency of musical instrument sounds, that is, acoustic features of most musical instruments vary according to the pitch (fundamental frequency, F0). To cope with this difficulty, we propose an F0-dependent multivariate normal distribution, where each element of the mean vector is represented by a function of F0. Our method first extracts 129 features (e.g., the spectral centroid, the gradient of the straight line approximating the power envelope) from a musical instrument sound and then reduces the dimensionality of the feature space into 18 dimension. In the 18-dimensional feature space, it calculates an F0-dependent mean function and an F0-normalized covariance, and finally applies the Bayes decision rule. Experimental results of identifying 6,247 solo tones of 19 musical instruments shows that the proposed method improved the recognition rate from 75.73% to 79.73%.

The perception of size in musical instrument sounds

There is size information in natural sounds. For example, as humans grow in height their vocal tract increases in length, and this produces a predictable decrease in formant frequency. Recent studies have shown that listeners can judge the relative size of two individuals with considerable precision, and they can recognize vowels scaled well beyond the range normally experienced. This paper reports two psychophysical studies designed to extend the research to musical instrument sounds. The first showed that listeners can discriminate the relative size of instruments reliably, although not quite as well as for voices. The second showed that listeners can recognize instrument sounds scaled in size and pitch well beyond the range of normal experience. The research supports the hypothesis that the auditory system applies some kind of active normalization to all input sounds. [Work supported by the U.K. Medical Research Council (G9901257, G9900362), and ONRIFO (Grant N00014-03-1-1023).]

Effect of critical-band smoothing of musical instrument spectral data

It has been found that second-order harmonic smoothing of musical instrument spectral data can have a significant effect on timbral perception, depending on the instrument tested [McAdams et al., J. Acoust. Soc. Am. 102, 882–897 (1999)]. With critical-band smoothing, the lower harmonics, since they are in different critical bands, retain their individual amplitudes and temporal envelopes. Thus, it is hypothesized that critical-band smoothing has a lesser perceptual effect on most instrument tones than harmonic smoothing. On the other hand, upper critical bands consist of groups of harmonics. It is hypothesized that it is difficult to hear out individual harmonics within critical bands. Thus, for each band the independent harmonic temporal envelopes can be replaced by a composite rms-amplitude envelope. Spectra within bands can be replaced by time-averaged spectra. Alternatively, time-dependent amplitude versus Bark-frequency spectral envelopes can be smoothed for each individual analysis frame. Further, amplitudes can be averaged in dB or linear units. Results for various processing combinations and various musical instrument sounds will be given and demonstrated.

Automatic Recognition of Isolated Monophonic Musical Instrument Sounds using kNNC

The instrument recognition system described in this paper classifies isolated monophonic musical instrument sounds using six features: cepstral coefficients, constant Q transform frequency spectrum, multidimensional scaling analysis trajectories, RMS amplitude envelope, spectral centroid and vibrato. Sounds from nineteen instruments of definite pitch, covering the note range C3-C6 and representing the major musical instrument families and subfamilies were used to test the system. Nearest neighbor classification was utilised and results were evaluated in terms of accuracy and reliability. Using the leave-one-out test strategy yielded an accuracy of 93% in instrument recognition, 97% in instrument family recognition, and 100% for sustain/impulsive instruments.

Detection of random alterations to time-varying musical instrument spectra.

The time-varying spectra of eight musical instrument sounds were randomly altered by a time-invariant process to determine how detection of spectral alteration varies with degree of alteration, instrument, musical experience, and spectral variation. Sounds were resynthesized with centroids equalized to the original sounds, with frequencies harmonically flattened, and with average spectral error levels of 8%, 16%, 24%, 32%, and 48%. Listeners were asked to discriminate the randomly altered sounds from reference sounds resynthesized from the original data. For all eight instruments, discrimination was very good for the 32% and 48% error levels, moderate for the 16% and 24% error levels, and poor for the 8% error levels. When the error levels were 16%, 24%, and 32%, the scores of musically experienced listeners were found to be significantly better than the scores of listeners with no musical experience. Also, in this same error level range, discrimination was significantly affected by the instrument tested. For error levels of 16% and 24%, discrimination scores were significantly, but negatively correlated with measures of spectral incoherence and normalized centroid deviation on unaltered instrument spectra, suggesting that the presence of dynamic spectral variations tends to increase the difficulty of detecting spectral alterations. Correlation between discrimination and a measure of spectral irregularity was comparatively low.

Behavioral and neural correlates of perceived and imagined musical timbre

The generality of findings implicating secondary auditory areas in auditory imagery was tested by using a timbre imagery task with fMRI. Another aim was to test whether activity in supplementary motor area (SMA) seen in prior studies might have been related to subvocalization. Participants with moderate musical background were scanned while making similarity judgments about the timbre of heard or imagined musical instrument sounds. The critical control condition was a visual imagery task. The pattern of judgments in perceived and imagined conditions was similar, suggesting that perception and imagery access similar cognitive representations of timbre. As expected, judgments of heard timbres, relative to the visual imagery control, activated primary and secondary auditory areas with some right-sided asymmetry. Timbre imagery also activated secondary auditory areas relative to the visual imagery control, although less strongly, in accord with previous data. Significant overlap was observed in these regions between perceptual and imagery conditions. Because the visual control task resulted in deactivation of auditory areas relative to a silent baseline, we interpret the timbre imagery effect as a reversal of that deactivation. Despite the lack of an obvious subvocalization component to timbre imagery, some activity in SMA was observed, suggesting that SMA may have a more general role in imagery beyond any motor component.

Musical Skill in Dementia: A Violinist Presumed to Have Alzheimer’s Disease Learns to Play a New Song

Previous studies have described patients with possible or probable Alzheimer’s disease (AD) who continued to play familiar songs skillfully, despite their dementias. There are no reports about patients with dementia who successfully learned to play new songs, and two papers describe failures of patients with AD to learn to play a new song although they continued to play familiar songs competently. In the present paper we describe a moderately demented patient (SL) with probable AD who learned to play a song (Cossackaya!) on the violin that was published after the apparent onset of his dementia. He showed modest retention of the song at delays of 0 and 10 minutes. This contrasts with his profound disturbance in both recall and recognition on other anterograde memory tests (word lists, stories, figures, environmental sounds, sounds of musical instruments), and marked impairment on measures of remote memory (famous faces, autobiographical memory). SL showed milder deficits in confrontation naming, verbal fluency and attention, but no dyspraxia or aphasic comprehension deficits. Except for the Block Design test, his visuospatial skills were intact. SL’s learning of the new song in the absence of any evidence of episodic memory is reminiscent of patients with temporal lobe amnesia who show better memory for song melody than for lyrics or verse, although his retention was not as good.