Chapter 7 Exercises: Timbre, Waveforms & Fourier's Revelation

These exercises are organized into five parts: conceptual foundations, mathematical application, spectral analysis and code, connections to physics and technology, and philosophical/creative engagement.

Part A: Definitions and Core Concepts

1. Define the following terms precisely and explain what role each plays in understanding timbre: - Waveform - Spectrum - Spectral envelope - Spectrogram - Time domain vs. frequency domain

Give a concrete example of a situation where the time domain representation is more useful, and a situation where the frequency domain representation is more useful.

2. Explain Fourier's theorem in plain language. Your explanation should: - State what the theorem claims (what can be decomposed, into what) - Explain why this is surprising or non-obvious - Describe what "infinitely many" sine waves means in practice - Give an example of a waveform that requires many sine waves to represent and explain why

3. What is the difference between the source spectrum and the spectral envelope of an instrument? Give two concrete examples showing how this distinction explains differences between instruments in the same family (e.g., two woodwinds, or two string instruments played in different ways).

4. Ohm's Acoustic Law states that the ear responds to the amplitudes of frequency components but not their phases (for sustained tones). What does this mean in practical terms? Describe one situation where phase matters to perception and one where it does not, according to current evidence.

5. What is a spectral centroid, and what perceptual quality does it correlate with? Give three examples from the Spotify spectral dataset (from section 7.8) of genres with high, medium, and low spectral centroids, and briefly explain why the instrumentation of each genre would produce those centroid values.

Part B: Mathematics of the Fourier Transform

6. A complex tone consists of the following sine wave components: - 100 Hz at amplitude 1.0 - 200 Hz at amplitude 0.5 - 300 Hz at amplitude 0.33 - 400 Hz at amplitude 0.25

a. What is the fundamental frequency of this tone? b. What musical interval does the 200 Hz component form above the 100 Hz component? c. What is the spectral centroid of this combination? (Hint: calculate the weighted average of the frequencies, weighted by amplitude.) d. If the 300 Hz component is removed, how does the spectral centroid change? Calculate the new value. e. Which component removal would produce the largest perceptual change in brightness? Justify your answer.

7. Explain the time-frequency uncertainty principle (the acoustic analog of Heisenberg's uncertainty principle) in the context of spectrograms. Specifically: - What are the two quantities that cannot simultaneously be measured with high precision? - How does window size affect this trade-off? - Why does this matter for analyzing music that includes both fast transients and slow harmonic content? - How do wavelets address this limitation?

8. The Constant-Q Transform (CQT) is described in section 7.12 as spacing frequency bins logarithmically rather than linearly. a. Why is logarithmic frequency spacing more natural for music than linear spacing? (Hint: think about the equal-tempered scale.) b. If a standard FFT with 4096 points and a 44100 Hz sample rate has a frequency resolution of about 10.8 Hz, what is the spacing between frequency bins in Hz? c. At 440 Hz (A4), the next semitone (A#4/Bb4) is about 26 Hz higher. Would the standard FFT bins be able to reliably distinguish these two pitches? Show your calculation. d. Would the CQT, which allocates more bins near lower frequencies, solve this problem? Explain.

9. Consider the phase relationship question from section 7.10. Two sounds are identical in spectral content (same frequencies, same amplitudes) but differ in the phase of every harmonic: - Sound A: all harmonics start at the same phase (a cosine wave, peaking at time zero) - Sound B: all harmonics start at random phases

a. What will Sound A's waveform look like in the time domain? b. What will Sound B's waveform look like? c. According to Ohm's Acoustic Law, should they sound identical? d. In practice, does phase difference between harmonics affect perceived timbre? What does the evidence say?

10. Explain what the Fast Fourier Transform (FFT) is, how it differs from the discrete Fourier transform (DFT), and why its development in 1965 was practically significant. Your answer should: - Describe the computational complexity improvement (n² vs. n log n) - Give a concrete example of a signal length and how much faster FFT is compared to direct DFT computation - List two fields outside music where the FFT's speed was decisive for technological development

Part C: Spectral Analysis and Code

11. Using the code in code/fourier_analysis.py as a reference (or running it yourself): a. What is the relationship between the number of FFT points (n) and the frequency resolution of the resulting spectrum? b. What is the relationship between the sample rate and the highest frequency that can appear in the FFT output (the Nyquist frequency)? c. If you analyze a 0.5-second recording sampled at 44100 Hz using a full-length FFT, what is the frequency resolution in Hz? d. What is the Nyquist frequency for standard CD quality (44100 Hz sampling)?

12. Examine the timbre_comparison.py code. The clarinet profile has much weaker even harmonics than odd harmonics. a. What physical principle of the clarinet's tube geometry causes this? b. How would the waveform of the clarinet (in the time domain) visually differ from the violin waveform because of this odd-harmonic emphasis? c. If you were to synthesize a clarinet-like tone and then tune it to A4 (440 Hz), what would the frequencies of its five strongest harmonics be?

13. Aiko Tanaka's analysis in section 7.7 revealed combination tones in the Bach chorale recording. a. If two vocal partials are at 880 Hz and 1100 Hz (a perfect fifth above the octave), what is the difference tone? What musical note does it correspond to? b. What is the cubic difference tone (2f₁ - f₂)? c. If the two voices are very slightly mistuned so that one partial is at 882 Hz instead of 880 Hz, at what rate will you hear amplitude beating? What is the perceptual effect of 2 Hz beating?

14. Spectral rolloff is the frequency below which 85% of a signal's total spectral energy is concentrated. Using the conceptual framework from this chapter (you do not need to run code): a. Would you expect a flute's spectral rolloff to be higher or lower than an oboe's at the same pitch and loudness? Explain. b. For a trumpet playing fortissimo vs. the same trumpet playing pianissimo at the same pitch, which has the higher spectral rolloff? Why? c. How does spectral rolloff relate to spectral centroid? Are they measuring the same thing?

15. The "chorus effect" in choral music (section 7.7) arises because multiple voices singing the same pitch with small tuning differences produce a richer, fuller sound than a single voice amplified to the same level. a. Using what you know about beating and combination tones, explain the physical mechanism of the chorus effect. b. Digital audio processors replicate the chorus effect by delaying and slightly pitch-shifting copies of a single audio signal. How does this approximate the physical mechanism? c. What is missing in the digital simulation compared to an actual choir in a reverberant space?

Part D: Physics Connections and Technology

16. Section 7.11 describes how MRI machines use the Fourier transform. Explain in general terms how: a. Spatial information about the body is encoded in the frequency content of the received radio signal b. The inverse Fourier transform is used to reconstruct a spatial image from frequency-domain data c. This process is analogous to — and different from — reconstructing a musical tone's harmonic content from its waveform

17. Heisenberg's uncertainty principle in quantum mechanics states that the more precisely you know a particle's position, the less precisely you can know its momentum, and vice versa. The chapter states that this is "a direct consequence of the Fourier relationship between position-space and momentum-space."

Explain the connection: a. In acoustics, what are the two "conjugate" quantities that cannot both be precisely known at the same time? b. In quantum mechanics, what are the two conjugate quantities in the uncertainty principle? c. Both cases involve a Fourier transform between two representations. Why does Fourier mathematics necessarily produce this kind of trade-off?

18. The phase vocoder (section 7.10) allows audio to be time-stretched or pitch-shifted independently of each other. Before the phase vocoder, these operations were coupled: slowing down a recording on tape also lowered its pitch. a. Why are time and pitch coupled in the tape domain? b. How does working in the Fourier (frequency) domain break this coupling? c. What audible artifacts can occur when time-stretching or pitch-shifting is pushed to extremes, and what causes them physically?

19. MP3 compression uses a model of human auditory masking — the phenomenon where a loud sound makes a nearby quieter sound inaudible. This allows the compression algorithm to discard audio data that the ear could not hear anyway. a. In spectral terms, what is "masking"? How does a loud harmonic at 1 kHz affect the perception of a quieter component at 1.05 kHz? b. How does this relate to the concept of critical bands in psychoacoustics? c. At very high compression ratios, what artifacts (audible distortions) does MP3 compression produce, and how do these relate to the discarding of spectral information?

20. The theremin is mentioned in Case Study 7.2. It produces nearly pure sine waves — single-frequency tones with very few harmonics. a. What would the spectrogram of a theremin melody look like compared to a violin melody at the same pitches? b. Why would a theremin spectrogram make it easy to see pitch glides (continuous transitions between pitches) that might be harder to read on a violin spectrogram? c. The theremin's pure sine wave timbre is often described as "otherworldly" or "eerie." Using the spectral centroid concept, explain why a pure sine wave sounds distinctly different from most acoustic instruments.

Part E: Aiko's Experiment, Emergence, and Philosophy

21. Aiko's Fourier analysis of the Bach motet in section 7.7 revealed emergent phenomena — combination tones and beating patterns that were not present in any single voice but arose from their interaction. Write a short essay (400–600 words) responding to the following:

"Aiko could explain every acoustic phenomenon she observed in terms of physical mechanisms — beating, combination tones, acoustic nonlinearity. Does this mean her experience of the choir was 'fully explained' by the physics? Or does the experience of the music contain something that the physical explanation leaves out?"

Your essay should take a clear position and engage with specific details from the chapter.

22. The Spotify spectral dataset shows systematic differences between genres in spectral centroid. Consider what is not captured by spectral statistics: a. Name three musical features that distinguish two genres with the same spectral centroid b. Could two tracks with identical spectrograms (same time-frequency-amplitude pattern) belong to different genres? If so, what would account for the genre difference? c. What does your answer tell us about the relationship between acoustic physics and musical meaning?

23. Fourier's theorem states that any periodic waveform can be represented as a sum of sine waves. But musical sounds are not perfectly periodic — they evolve over time, with subtle continuous variations in pitch, timbre, and amplitude. a. Strictly speaking, does Fourier's theorem apply to music? b. How does the spectrogram (windowed Fourier transform) address this limitation? c. What does this tell us about the relationship between mathematical models and physical reality? Is the Fourier framework "true" for music, or is it an approximation?

24. Theme 4 of this textbook is "Technology as Mediator." Reflect on the role of the Fourier transform and spectrogram technology as a mediator between music and physics: a. What aspects of music were made newly visible by spectrogram technology? b. What aspects of music remain invisible to the spectrogram? c. Has the existence of spectral analysis technology changed how musicians, composers, or producers make music? Give a specific example. d. Is this technology "neutral" — simply a mirror of musical reality — or does it shape what we look for and what we value in music?

25. Design Challenge: You are designing a music-education tool for children aged 8–12 that uses real-time spectrogram visualization to teach about timbre and the harmonic series.

Describe your tool in 400–600 words: - What hardware and software would it require? - What visual representation would you use, and why? (What are the limitations of a standard spectrogram for children?) - What activities would you design around it to teach the concept that different instruments have different spectral profiles? - How would you address the challenge that the standard spectrogram can be confusing even for adults? - What would you measure to evaluate whether your tool is effective?

End of Chapter 7 Exercises