AI Music Glossary
Key terms in AI music generation, copyright, and synthetic media. Plain definitions, no jargon.
Audio Inpainting
A technique that regenerates a selected portion of an audio track while keeping the rest intact. Used for fixing artifacts or refining specific sections of AI-generated music.
Content ID
YouTube's automated system for identifying copyrighted content. Scans uploaded audio against a database of reference files. Can flag AI-generated music that closely resembles copyrighted recordings.
Deepfake Audio
AI-generated audio designed to replicate a specific person's voice or musical style with high fidelity. Raises significant legal and ethical concerns when used without consent.
Diffusion Model
A type of generative AI that creates data (including audio) by gradually removing noise from a random signal. Many modern AI music tools use diffusion-based architectures.
Fair Use
A US legal doctrine allowing limited use of copyrighted material without permission for purposes like commentary, criticism, or transformation. Central to the legal debate over AI training on copyrighted music.
Generative Audio
AI-produced sound, including music, speech, and sound effects. Encompasses text-to-song, text-to-speech, and other audio synthesis methods.
Latent Space
A compressed mathematical representation of data learned by an AI model. In music generation, the latent space encodes musical features (rhythm, harmony, timbre) as numerical vectors.
MIDI
Musical Instrument Digital Interface. A protocol for communicating musical information (notes, velocity, timing) between devices and software. Some AI tools output MIDI for further editing in DAWs.
Prompt Engineering
The practice of crafting text inputs to guide AI output. In AI music, effective prompts specify genre, mood, instrumentation, tempo, and style to improve generation quality.
Stem Separation
The process of isolating individual audio tracks (vocals, drums, bass, instruments) from a mixed recording. AI has significantly improved stem separation quality since 2023.
Style Transfer
Applying the musical characteristics of one piece (genre, arrangement, production style) to guide the generation of a new piece. Distinct from copying a specific recording.
Synthetic Media
Content (audio, video, images, text) generated or substantially modified by AI. AI-generated music is a subcategory of synthetic media.
Text-to-Song
AI capability that generates complete songs (vocals, instrumentation, arrangement) from text descriptions. The primary feature of tools like Suno and Udio.
Training Data
The dataset used to train an AI model. For music models, this typically includes recorded songs, MIDI files, and audio samples. The legality of using copyrighted music as training data is the central legal question in AI music.
Watermarking (Audio)
Embedding imperceptible identifying information in audio files. Can be used to identify AI-generated content, track provenance, or assert ownership. Not all AI music tools apply watermarks.