AI Music Extension and Continuation Quiz

Answers and explanations appear as you take the quiz.

1. How does an autoregressive model like MusicGen continue an audio clip?
   • It reverses the clip and plays it backwards
   • It searches a library for a matching recording
   • It predicts further codec tokens after the tokens of the supplied audio
   • It stretches the clip to a longer duration
2. Which neural codec does MusicGen use to turn audio into tokens?
   • EnCodec
   • MP3
   • Demucs
   • MIDI
3. What does audio inpainting do?
   • Adds reverb to the whole track
   • Regenerates a masked region in the middle while preserving audio on both sides
   • Converts audio into sheet music
   • Separates vocals from instruments
4. Why are diffusion models well suited to inpainting?
   • They generate audio one note at a time
   • They require no training data
   • They only work on silence
   • The known audio can be held fixed while only the masked span is denoised
5. In a remix or audio-to-audio mode, what does the strength setting control?
   • The output volume
   • The file format
   • How much of the original input remains in the result
   • The number of stems produced
6. What does a source separation model such as Demucs do?
   • Splits a mixed track into stems like vocals, drums, bass and other
   • Generates lyrics from a melody
   • Detects copyright infringement
   • Compresses audio for streaming
7. Why can a motif from the first minute be forgotten in a long extension?
   • The model deletes the original file
   • Codec tokens expire after a minute
   • Extensions always switch genre
   • The model only sees a limited context window, often via a sliding window of recent audio
8. What is the rights situation when you extend a commercial recording you do not own?
   • The extension is automatically yours alone
   • You create a derivative of that recording, and many services forbid uploading it
   • It becomes public domain
   • Rights disappear after 30 seconds of new audio