Guide-linked quiz · Medium Level

Musical Timbre Transfer Quiz

Review timbre-transfer goals, pitch and loudness controls, differentiable synthesis, harmonic-noise components, and spectral reconstruction loss.

Related guide pathsMusical Timbre Transfer
Question 1 of 8

In timbre transfer, what is preserved from the original audio?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. In timbre transfer, what is preserved from the original audio?
    • The instrument's spectral fingerprint
    • The recording's file format
    • The pitch, loudness, and timing of the performance
    • The background reverb only
  2. What does 'timbre' actually refer to?
    • How loud a sound is
    • The tonal color that distinguishes instruments playing the same note
    • The speed of a song
    • The key signature
  3. What makes Google's DDSP approach especially data-efficient?
    • It bakes classic differentiable synthesizer structure into the network
    • It uses enormous labeled datasets
    • It ignores pitch entirely
    • It only works on drums
  4. Why must the synthesizer in DDSP be 'differentiable'?
    • So it can run on mobile phones
    • So it can play multiple notes
    • So it compresses audio
    • So gradients can flow back through it during training
  5. Which two control signals does DDSP typically extract from the input performance?
    • Tempo and key
    • Fundamental frequency and loudness envelope
    • File size and bitrate
    • Lyrics and chords
  6. What source type did the original DDSP paper use in the harmonic-plus-noise experiments discussed here?
    • Every possible polyphonic mixture without any representation changes
    • Only silent recordings with no pitched content
    • Monophonic sources modeled around a single fundamental-frequency trajectory
    • Only MIDI event labels with no audio sources
  7. In the DDSP approach described here, which synthesis components render the predicted timbre controls into sound?
    • Only a text-to-speech phoneme decoder
    • Only a classifier assigning an instrument name
    • A nearest-neighbor lookup of complete prerecorded performances
    • A harmonic oscillator bank combined with filtered noise
  8. What does DDSP's spectral reconstruction objective compare during timbre-transfer training?
    • The generated and target audio's spectrogram representations
    • Only the two recordings' file names
    • Only the number of notes written in sheet music
    • The album covers' color histograms

Keep testing yourself

More quizzes picked for your level.

View all quizzes