Acoustic Scene Classification Quiz

Answers and explanations appear as you take the quiz.

1. How does acoustic scene classification differ from sound event detection?
   • They are exactly the same task
   • ASC identifies the overall environment, while event detection spots specific sounds
   • ASC only works on music
   • Event detection ignores time entirely
2. What is the 'device mismatch' problem in ASC?
   • Speakers playing audio too loudly
   • Files saved in the wrong format
   • A model trained on one microphone performing poorly on another
   • Two scenes sounding identical
3. Which input representation is standard for ASC models?
   • Raw text transcripts
   • JPEG images of the location
   • GPS coordinates
   • Log-mel spectrograms
4. Why do ASC models pool features over many seconds rather than single moments?
   • Because scenes are defined by long-term acoustic statistics, not single events
   • To reduce file size
   • Because microphones are slow
   • To remove all noise
5. Which research challenge has driven much of the progress in ASC?
   • The ImageNet competition
   • The Loebner Prize
   • The Turing Test
   • The DCASE Challenge
6. What inference-computation budget did the DCASE 2024 low-complexity acoustic-scene task specify for a one-second audio clip?
   • Exactly 30 billion operations for every device
   • At most 30 million multiply-accumulate operations
   • No computation limit of any kind
   • A limit measured only by the recording filename length
7. How can quantization help an acoustic-scene model meet a parameter-storage budget?
   • By storing weights with fewer bits per parameter
   • By adding more high-precision copies of every weight
   • By replacing the memory budget with an accuracy score
   • By making the microphone sample rate unlimited
8. What variation is device-aware augmentation intended to simulate for acoustic-scene training?
   • Only changes in the spelling of scene labels
   • Only a different album-cover image
   • Changes in recording-device characteristics such as microphone frequency response
   • A guarantee that all environments have identical acoustics