Riffusion Spectrogram Diffusion Quiz

Answers and explanations appear as you take the quiz.

1. What existing AI model did Riffusion fine-tune to generate music?
   • Stable Diffusion, an image generation model
   • A speech recognition model
   • A large language model
   • A reinforcement learning agent
2. On a Riffusion spectrogram image, which direction typically represents increasing time?
   • Bottom to top along the vertical axis
   • Left to right along the horizontal axis
   • From the center outward
   • Time is not represented
3. Why does Riffusion need an algorithm like Griffin-Lim?
   • To translate the text prompt into English
   • To compress the final audio file
   • To reconstruct phase information missing from the spectrogram
   • To detect the song's genre
4. What can prompt interpolation do in Riffusion?
   • Turn generated music into a transcript
   • Separate instruments into stems
   • Replace the image model with a language model
   • Blend between two prompt-conditioned sound ideas
5. How was Riffusion originally created and released?
   • As a hobby project by two creators in 2022
   • As a government research grant
   • As a paid product by a major record label
   • As part of a university PhD thesis only
6. Besides prompts, what additional conditioning does Riffusion’s core repository list for its diffusion pipeline?
   • A speaker diarization label
   • Image conditioning
   • A MIDI transcription
   • A microphone array geometry
7. In a Riffusion spectrogram, what does pixel intensity represent?
   • The lyric transcript confidence
   • The stereo microphone location
   • The magnitude or energy of a time-frequency component
   • The number of diffusion steps alone
8. What does a mel-frequency axis do compared with uniform spacing in hertz?
   • It converts frequency into spoken words
   • It makes every frequency bin equally spaced in hertz
   • It stores only stereo position
   • It uses a nonlinear spacing intended to better reflect perceived pitch differences