the mel scale is: a nonlinear frequency scale that models human pitch perception

mel-frequency cepstral coefficients (MFCCs) capture: speech features on a perceptual scale

Mel-frequency cepstral coefficients (MFCCs) represent sound on a perceptual scale

Rate-distortion theory: minimum bits to represent data within distortion D

How many bits do we need to perfectly copy a song?

Pre-LN

Why is normalizing data like tuning instruments before a concert?

Fisher information

Fisher information measures information about unknown parameters

AI content watermarking

AI content watermarking embeds imperceptible signals

WordPiece tokenization does: similar to BPE but uses likelihood instead of frequency

WordPiece tokenization splits words into subwords based on token likelihood rather than frequency