In many ABX tests and FFT/IMD measurements, as here in ASR, we ensure two amps or DACs have no measurable noise or distortion in the 20 Hz–20 kHz band. Yet one device often shows extra noise above 20 kHz, which we typically assume is inaudible. My question is: Has anyone done a double‑blind study measuring “emotional engagement” (rather than just detection of tone differences) when two devices differ only in ultrasonic noise? In other words, Device A and Device B both show zero measurable noise in 20 Hz–20 kHz, but Device B has, say, a 100 kHz spur at –60 dBFS while Device A does not. Can listeners still feel a difference in how “emotionally satisfying” music sounds, even though neither spur is audible in isolation (and neither spur can be reliably identified by listeners in a double-blind test as has already been proven numerous times)? In other words, even if our ears and conscious brain cannot pick out any differences between the two “audibly transparent” devices, can our subconscious brain still perceive the differences and thereby affect our level of listening pleasure of the two devices differently? If anyone knows of any peer‑reviewed or AES‑conference papers on this topic, or any informal but well‑controlled experiments, I’d be grateful for references. If not, I’ve sketched a detailed study design below; I welcome any feedback.
1. Study Overview
1.1 Working Title
“Silent but Felt: A Double-Blind Study of Ultrasonic Noise’s Impact on Emotional Engagement in Music Playback”
2. Study Objectives
2.1 Primary Goal
Determine if ultrasonic noise (≥ 20 kHz) alters emotional engagement with music.
2.2 Secondary Goals
2.2.1 Identify susceptibility of various musical passages to ultrasonic effects.
2.2.2 Quantify the minimum ultrasonic noise level (in dBFS) for noticeable engagement changes.
3. Device Selection and Characterization
3.1 Device Selection
3.1.1 Device A: Modified to minimize ultrasonic noise.
3.1.2 Device B: Same model with intentional ultrasonic spur (e.g., –60 dBFS at 100 kHz).
3.2 Key Constraint
Ensure identical in-band (20 Hz–20 kHz) measurements.
3.3 In-Band Equivalence Confirmation
3.3.1 FFT Analysis: Compare noise floors.
3.3.2 IMD Test: Verify no intermodulation into the audible band.
3.3.3 A-Weighted Noise Floor: Ensure both devices < –90 dBFS.
4. Listener Recruitment & Screening
4.1 Sample Size and Profile
4.1.1 Recruit ≥ 30 participants, including critical and casual listeners.
4.1.2 Confirm normal hearing (≤ 20 dB HL).
4.2 Screening Procedures
4.2.1 Basic Audiometric Check.
4.2.2 Engagement Consistency Test with repeated musical excerpts.
5. Music Selection (Stimuli)
5.1 Excerpts (60–90 seconds each)
5.1.1 Solo Piano (e.g., "Clair de Lune")
5.1.2 Solo Violin/Cello (e.g., Bach Suite)
5.1.3 Small Ensemble (e.g., Jazz Trio)
5.1.4 Full Orchestra (e.g., "Boléro")
5.1.5 Vocal (e.g., Soprano Aria)
5.1.6 Acoustic Guitar + Percussion
5.2 Audio Quality
Ensure 96 kHz/24-bit or higher, normalized peak levels.
6. Listening Environment & Equipment Setup
6.1 Room Requirements
6.1.1 Acoustically treated room (RT60 ≤ 0.3 s).
6.1.2 Standardized speaker placement.
6.2 Equipment Setup
6.2.1 High-quality DAC with fixed volume.
6.2.2 Silent, relay-based A/B switch for device toggling.
7. Blinding and Randomization
7.1 Device Labeling
Label as "Device 1" and "Device 2" (blind to participants and operators).
7.2 Randomization and Trials
7.2.1 Random device order per trial.
7.2.2 Include dummy trials to measure bias.
7.2.3 Present each track under A→B, B→A, and dummy conditions.
8. Experimental Procedure
8.1 Calibration & Familiarization
8.1.1 SPL Calibration using 1 kHz tone.
8.1.2 Practice trials without ratings.
8.2 Listening Trials
8.2.1 First Playback.
8.2.2 30-second Pause.
8.2.3 Second Playback.
8.2.4 Rating Form completion.
8.3 Breaks & Fatigue Management
8.3.1 Breaks after every 6 trials.
8.3.2 Total session time ~1 hour 50 minutes.
9. Data Collection & Analysis
9.1 Primary Endpoint
9.1.1 Binary choice accuracy analyzed via binomial tests.
9.2 Secondary Endpoint
9.2.1 Engagement rating differences analyzed using paired t-tests.
9.3 Track-Specific Analysis
9.3.1 Sensitivity evaluation across different musical contexts.
10. Controlling Confounds
10.1 Volume Matching
Verify ±0.01 dB matching.
10.2 Equipment Consistency
Ensure identical cables and consistent setup.
10.3 Listener Blinding
Double-blind procedures for all participants.
10.4 Environmental Controls
Maintain controlled room conditions.
11. Reporting & Interpretation of Results
11.1 Data Presentation
11.1.1 FFT plots (0–200 kHz) and IMD results.
11.1.2 Accuracy and rating difference tables.
11.1.3 Track-specific performance analysis.
11.2 Qualitative Feedback
Summarize participant comments.
11.3 Conclusions
11.3.1 Assess significance of differences.
11.3.2 Determine ultrasonic noise impact on emotional engagement.
12. Study Robustness
12.1 Focus
Concentrate solely on ultrasonic effects.
12.2 Design
Double-blind, placebo-controlled structure.
12.3 Metrics
Use emotional engagement as the primary endpoint.
12.4 Statistical Power
Sample size and rigorous analysis ensure statistical significance.
12.5 Statistical Rigor
Binomial tests for choice accuracy and paired t‑tests (or Wilcoxon) for rating differences ensure robust inference.
1. Study Overview
1.1 Working Title
“Silent but Felt: A Double-Blind Study of Ultrasonic Noise’s Impact on Emotional Engagement in Music Playback”
2. Study Objectives
2.1 Primary Goal
Determine if ultrasonic noise (≥ 20 kHz) alters emotional engagement with music.
2.2 Secondary Goals
2.2.1 Identify susceptibility of various musical passages to ultrasonic effects.
2.2.2 Quantify the minimum ultrasonic noise level (in dBFS) for noticeable engagement changes.
3. Device Selection and Characterization
3.1 Device Selection
3.1.1 Device A: Modified to minimize ultrasonic noise.
3.1.2 Device B: Same model with intentional ultrasonic spur (e.g., –60 dBFS at 100 kHz).
3.2 Key Constraint
Ensure identical in-band (20 Hz–20 kHz) measurements.
3.3 In-Band Equivalence Confirmation
3.3.1 FFT Analysis: Compare noise floors.
3.3.2 IMD Test: Verify no intermodulation into the audible band.
3.3.3 A-Weighted Noise Floor: Ensure both devices < –90 dBFS.
4. Listener Recruitment & Screening
4.1 Sample Size and Profile
4.1.1 Recruit ≥ 30 participants, including critical and casual listeners.
4.1.2 Confirm normal hearing (≤ 20 dB HL).
4.2 Screening Procedures
4.2.1 Basic Audiometric Check.
4.2.2 Engagement Consistency Test with repeated musical excerpts.
5. Music Selection (Stimuli)
5.1 Excerpts (60–90 seconds each)
5.1.1 Solo Piano (e.g., "Clair de Lune")
5.1.2 Solo Violin/Cello (e.g., Bach Suite)
5.1.3 Small Ensemble (e.g., Jazz Trio)
5.1.4 Full Orchestra (e.g., "Boléro")
5.1.5 Vocal (e.g., Soprano Aria)
5.1.6 Acoustic Guitar + Percussion
5.2 Audio Quality
Ensure 96 kHz/24-bit or higher, normalized peak levels.
6. Listening Environment & Equipment Setup
6.1 Room Requirements
6.1.1 Acoustically treated room (RT60 ≤ 0.3 s).
6.1.2 Standardized speaker placement.
6.2 Equipment Setup
6.2.1 High-quality DAC with fixed volume.
6.2.2 Silent, relay-based A/B switch for device toggling.
7. Blinding and Randomization
7.1 Device Labeling
Label as "Device 1" and "Device 2" (blind to participants and operators).
7.2 Randomization and Trials
7.2.1 Random device order per trial.
7.2.2 Include dummy trials to measure bias.
7.2.3 Present each track under A→B, B→A, and dummy conditions.
8. Experimental Procedure
8.1 Calibration & Familiarization
8.1.1 SPL Calibration using 1 kHz tone.
8.1.2 Practice trials without ratings.
8.2 Listening Trials
8.2.1 First Playback.
8.2.2 30-second Pause.
8.2.3 Second Playback.
8.2.4 Rating Form completion.
8.3 Breaks & Fatigue Management
8.3.1 Breaks after every 6 trials.
8.3.2 Total session time ~1 hour 50 minutes.
9. Data Collection & Analysis
9.1 Primary Endpoint
9.1.1 Binary choice accuracy analyzed via binomial tests.
9.2 Secondary Endpoint
9.2.1 Engagement rating differences analyzed using paired t-tests.
9.3 Track-Specific Analysis
9.3.1 Sensitivity evaluation across different musical contexts.
10. Controlling Confounds
10.1 Volume Matching
Verify ±0.01 dB matching.
10.2 Equipment Consistency
Ensure identical cables and consistent setup.
10.3 Listener Blinding
Double-blind procedures for all participants.
10.4 Environmental Controls
Maintain controlled room conditions.
11. Reporting & Interpretation of Results
11.1 Data Presentation
11.1.1 FFT plots (0–200 kHz) and IMD results.
11.1.2 Accuracy and rating difference tables.
11.1.3 Track-specific performance analysis.
11.2 Qualitative Feedback
Summarize participant comments.
11.3 Conclusions
11.3.1 Assess significance of differences.
11.3.2 Determine ultrasonic noise impact on emotional engagement.
12. Study Robustness
12.1 Focus
Concentrate solely on ultrasonic effects.
12.2 Design
Double-blind, placebo-controlled structure.
12.3 Metrics
Use emotional engagement as the primary endpoint.
12.4 Statistical Power
Sample size and rigorous analysis ensure statistical significance.
12.5 Statistical Rigor
Binomial tests for choice accuracy and paired t‑tests (or Wilcoxon) for rating differences ensure robust inference.
