Vibrato

literature review

1. Perceptual Evaluation of Vibrato Models

Vincent Verfaille, Catherine Guastavino, Philippe Depalle

Proceedings of the Conference on Interdisciplinary Musicology (CIM05)

• spectral envelope modulation
• frequency-dependent hysteresis behavior
• test: preference for sounds with modulated spectral envelope
• offers great literature review & background info on vibrato pitch, rate, extent, shape, and temporal evolution
• provides equations for modeling FM, AM, SEM pulsations
• vibrato is made of at least one of these three kind of modulations:

1. amplitude modulation (predominant in wind and brass instr)
2. frequency modulation (predominant in voice and string instr)
3. spectral envelope modulation and hysteresis (existing in wind, brass, voice)

• Questions & thoughts:

* How much of the analysis is instrument-specific, and how much is it performer/ context/ style-specific?

* It'd be interesting to see a more music-passage specific (higher level) analysis as opposed to instrument-specific modelling

2. An Investigation of Vocal Vibrato for Synthesis

Robert Maher, James Beauchamp

Applied Acoustics 30 (1990) 219-245

• focuses on vocal vibrato & presents a new panned-wavetable synthesis method
• role of spectrum modulation
• findings: "Inclusion of typical random fluctuations of vibrato rate, vibrato depth, and nominal sung frequency resulted in no quality preference over examples with constant values. Inclusion of vibrato-induced spectrum modulation resulted in a substantial improvement over examples having constant spectra."

3. Synthesis of Vibrato Singing

Yoram Meron and Keikichi Hirose

2000 IEEE International Conference on Acoustics

• matches the phase of the vibrato in the original unit (used for synthesis in the source database) with the vibarto phase of the synthesized (target) vibrato sound.
• useful section: 3.1 Vibrato detection

1. rate between 4.5 adn 7.0 Hz
2. modulation depth between 4% and 15%
3. extracted pitch - interpolated extrme points, compare to threshold

--> sequence of 5+ extreme points which pass all these tests are marked as vibrato anchor points

4. Automatic detection of vibrato in monophonic music

Hee-Suk Pang, Doe-Hyun Yoon

Pattern Recognition 38 (2005) 1135-1138

• method that automatically detects vibrato using (1) vibrato rate, (2) extent, and (3) intonation
• potential applications: wav-to-midi conversion
• uses maximum likelihood estimation (understand this..?)
• vibrato existence as product of (1) vibrato rate and (2) normalized vibrato extent probabilities
• Questions & thoughts:

• extend to polyphonic music with a vibrato solo-line? polyphonic music with overall vibrato quality?
• mentions how the recommended coefficient values can (should) be modified depending on musical style... It'd be nice to know how style affects these.

Where to go from here

• Goals
  • take a simple monophonic wav file and try to detect locations where vibrato is used
  • characterize parameters of vibratos for different style/ context
  • ultimately: try to apply expressive nuances/ vibratos to scores and MIDI files, etc. Try to impart characteristics of "style"
• Approach
  • avoid being instrument-specific
  • make it easily applicable to computer-generated sound, ensembles
  • aim for a higher level characterization