Spectral enrichment and wall losses in trombones played at high dynamic levels

John Chick, Shona Logie, Murray Campbell, Joël GilbertHAL 2012

TL;DR

This paper investigates how the "brassy" sound of a trombone at high volumes is created by the interplay of nonlinear wave distortion and energy loss due to friction with the tube walls. By experimenting with cylindrical tubes of varying diameters, the researchers found that while nonlinear distortion enriches the sound spectrum, wall losses can dampen these higher frequencies, creating a complex trade-off that influences the instrument's timbre.

What Problem Does This Paper Address?

The characteristic "brassy" sound of brass instruments, particularly at high playing dynamics, is a complex acoustic phenomenon. While it's understood that this sound arises from nonlinear effects within the instrument's bore, leading to spectral enrichment (more harmonics), the exact mechanisms and their interplay with other physical processes remain an area of active research. Specifically, the authors aim to disentangle the contributions of two key factors:

  1. Nonlinear Wave Steepening: As sound waves travel through the instrument, especially at high amplitudes, their waveforms can distort, leading to the generation of higher harmonics. This is analogous to a "shock wave" forming.
  2. Viscothermal Wall Losses: Friction and heat exchange between the sound wave and the inner walls of the instrument cause energy dissipation, particularly affecting higher frequencies.

The challenge is to understand how these two opposing effects – one creating higher frequencies and the other damping them – combine to produce the final radiated sound, and how this balance changes with the instrument's geometry, specifically its bore diameter. This is crucial for understanding and potentially modeling the timbre of brass instruments.

Methods

The researchers employed a combination of experimental measurements and computational simulations.

  • Experimental Setup: They used five stainless steel cylindrical tubes of varying diameters (6mm, 8mm, 10mm, 12mm, and 14mm) and a fixed length (2.96m). These diameters were chosen to represent a range of trombone bore sizes. A loudspeaker driver produced a sine wave input signal, and pressure transducers and microphones were strategically placed to measure:
  • Input pressure at the tube entrance.
  • Pressure at two points within the tube (near the entrance and near the middle).
  • Radiated sound pressure at a distance from the tube exit. The experiments were conducted at specific frequencies corresponding to the 13th resonant mode of each tube, chosen for its playability and to ensure sufficient excitation of nonlinear effects.

  • Computational Simulation: They used a "weakly nonlinear propagation theory" to model the sound propagation. This theory accounts for nonlinear distortion but treats viscothermal losses as linear. The simulation aimed to replicate the experimental conditions and provide a theoretical baseline for comparison.

  • Analysis Metrics:
    • Transfer Function (T14): The ratio of the radiated sound pressure to the input pressure, indicating how efficiently sound is transmitted through the tube and radiated.
  • Internal Transfer Function (T23): The ratio of pressure at the mid-point of the tube to the pressure at the entrance, used to isolate the effects of nonlinear distortion and wall losses within the tube, independent of end radiation effects.
  • Spectral Centroid: A measure of the "brightness" or average frequency content of the sound. An increase in spectral centroid indicates a shift towards higher frequencies.

Datasets

Not discussed in the paper. The paper presents experimental data collected by the authors and simulated data generated using their model.

Key Insights and Ideas

  • Conflicting Forces: The "brassy" sound is a result of a delicate balance: nonlinear distortion adds high frequencies, while wall losses remove them. The dominance of one over the other changes with tube diameter.
  • Diameter Matters: Larger diameter tubes experience less significant viscothermal losses relative to their volume, allowing nonlinear distortion to have a more pronounced effect on spectral enrichment.
  • Trombone-like Diameters: For diameters closer to that of a modern trombone (around 14mm), the experimental results showed a greater increase in spectral enrichment with increasing amplitude compared to simulations. This suggests that the simplified model might not fully capture the complex radiation dynamics in larger tubes.
  • Viscothermal Losses are Significant: Even in relatively large tubes, viscothermal losses play a crucial role. The simulation consistently under-predicted the amount of loss observed experimentally, particularly for internal transfer functions.
  • Small Tubes are "Lossy": The smallest tube (6mm) showed a significant damping of the transfer function, highlighting how viscothermal losses can dominate in narrow bores, making them less suitable for producing rich, brassy sounds.

Key Results

  • Transfer Function (T14): T14 generally increased with bore diameter, indicating more efficient sound radiation from larger tubes. This increase was more pronounced at higher input amplitudes, especially for the 14mm tube, suggesting a stronger nonlinear effect. Experimental results for T14 were generally higher than simulated results, particularly for larger diameters.
  • Internal Transfer Function (T23): This metric, which isolates internal effects, showed that simulations tended to over-predict T23 compared to experiments, suggesting the simulation might be underestimating viscothermal losses.
  • Spectral Centroid:
    • Simulated spectral centroid increased with decreasing tube diameter for high input amplitudes, peaking around 9-10mm.
  • Experimental spectral centroid also showed a peak around 10mm, but the absolute values were significantly higher than simulated values (up to 40% higher), indicating that the simulation was under-predicting the spectral enrichment.
  • Internal spectral centroid measurements (near the middle of the tube) were lower than far-field measurements, as expected, with values around 1.3.

Limitations and Future Work

  • Cylindrical Tubes vs. Real Instruments: The study used idealized cylindrical tubes, not the complex, evolving bore profiles of actual brass instruments. This simplification limits direct application to real instruments but serves as a foundational step.
  • Monopole Radiation Assumption: The simulation's assumption of monopole radiation from the tube exit may not accurately represent the realistic radiation field, especially for higher frequencies and larger tubes, contributing to discrepancies between experimental and simulated results.
  • Linear Viscothermal Loss Model: The simulation treated viscothermal losses as linear, whereas they can be amplitude-dependent.
  • Future Work: The authors suggest that future research should investigate the effects of more realistic bore profiles and explore more advanced simulation models that can better capture nonlinear viscothermal losses and radiation phenomena. Understanding the interplay of these factors in actual instruments is the next logical step.

Summary updated

Abstract

International audience
BibTeX
@inproceedings{chick2012spectral,
  author = {John Chick and Shona Logie and Murray Campbell and Joël Gilbert},
  title = {{Spectral enrichment and wall losses in trombones played at high dynamic levels}},
  booktitle = {HAL},
  year = {2012},
}

Loading related content...

Figure