Glossary · 56 terms
AI music glossary
The words you meet when you make music with AI, from the studio basics (stems, BPM, LUFS, sidechain) to the AI ones (inpainting, voice clones, reference tracks) and the names of Soundverse’s own tools. Each term has its own link to share.
Music basics
- Baralso Measure
- A bar is a short unit of musical time that groups a fixed number of beats, set by the time signature. In 4/4 each bar holds four beats, and song sections are counted in bars: an 8-bar verse, a 16-bar chorus.
- Beat
- The beat is the steady pulse you tap your foot to. Tempo says how fast beats come, and the time signature says how many make a bar.
- BPMalso Beats per minute
- BPM is a song's tempo: how many beats happen in one minute. A ballad might sit near 70 BPM, house music usually runs 120 to 130 BPM and drum and bass around 170. Soundverse's Tempo & Key Detector reads a track's BPM from the audio.
- Tempo & Key Detector
- Bridge
- A bridge is a contrasting section, usually heard once after the second chorus, that breaks the song's pattern before the final chorus.
- Chorus
- The chorus is the repeated section that carries a song's main message and usually its title. It is often the most memorable, highest-energy part.
- Hook
- A hook is a short musical or lyrical phrase built to stick in the listener's head, often in the chorus or the first seconds of a song.
- Key
- A song's key is its home note and scale, such as A minor or E-flat major. Parts in the same or a related key sound like they belong together, which is why DJs and producers match keys. Soundverse's Tempo & Key Detector finds the key from the audio.
- Tempo & Key Detector
- Loop
- A loop is a section of audio or MIDI, often one to eight bars long, repeated seamlessly to build a groove or a backing track.
- Melody
- The melody is the line of single notes a listener hums back, usually carried by the lead vocal or a lead instrument. Soundverse's Melody to Song Generator builds a whole song around a melody you hum, sing or play.
- Melody to Song Generator
- Tempo
- Tempo is the speed of a piece of music, measured in BPM. Changing the tempo of a recording without changing its pitch is called time-stretching.
- Time signature
- A time signature says how many beats are in each bar and which note value counts as one beat. Most popular music is in 4/4; a waltz is in 3/4.
- Verse
- A verse is a section whose lyrics change each time it comes round, telling the story between choruses while the melody stays largely the same.
Recording and production
- A cappellaalso Acapella
- An a cappella is a vocal with no instrumental backing. Producers use a cappellas for remixes and layering, and Soundverse's AI Singing Generator sings your lyrics as one.
- AI Singing Generator
- Arrangement
- The arrangement is which instruments play when, and the order and length of a song's sections. Arranging turns a sketch into a song with an intro, builds and an ending.
- Automation
- Automation is a recorded change to a setting over time, such as a fade, a filter sweep or more reverb on the last chorus. In Soundverse AI Studio you automate volume, pan and effect settings on the timeline.
- AI Studio
- DAWalso Digital audio workstation
- A DAW is software for recording, editing, arranging and mixing music on multiple tracks. An AI studio is a DAW with AI generation built into its timeline.
- What is an AI studio?
- MIDI
- MIDI is a format for musical instructions rather than sound: which note, when, for how long and how hard. A MIDI part can be played on any instrument and edited note by note, and Soundverse can turn a recording into MIDI or a MIDI melody into a sung song.
- Audio to MIDIMIDI to Song Generator
- Piano roll
- A piano roll is the grid editor for MIDI: pitch runs up the side, time runs across, and each note is a block you can move, stretch or delete.
- Sample
- A sample is a piece of existing audio, such as a drum hit, a loop or a phrase from a recording, reused in a new track. Clearing the rights to a sample before release is the producer's job.
- Stemsalso Stem
- Stems are the separate parts of a mix, such as vocals, drums, bass and other instruments, each as its own audio file. They let you remix, rebalance, practise along or hand a song to someone else to mix.
- Stem Separator
- Timeline
- The timeline is the main view of a project in a DAW, with time running left to right and each track stacked as a lane.
- Track
- A track is one lane on the timeline that holds a single part, such as a vocal, a drum loop or a MIDI synth line, with its own volume, pan and effects.
Mixing and mastering
- Compression
- Compression turns the loudest moments of a sound down so its level is more even, which helps a vocal or a drum hit sit steadily in a mix.
- De-esser
- A de-esser tames harsh "s" and "sh" sounds (sibilance) in a vocal by turning down only the frequencies where they live, only when they happen. Soundverse AI Studio's De-Esser One works on several sibilance bands at once.
- EQalso Equalization, Equaliser
- EQ boosts or cuts ranges of frequencies in a sound, like a more detailed bass and treble control. It is used to clear room for each instrument and to fix muddy or harsh tones.
- Limiter
- A limiter is a compressor with a hard ceiling: it stops a signal from going above a set level. It is usually the last step in mastering.
- LUFSalso Loudness units relative to full scale
- LUFS measures perceived loudness over time. Streaming services turn tracks up or down to a target, often around −14 LUFS, so mastering to a sensible target keeps a song from being turned down. Soundverse AI Studio's audio export offers Streaming −14, Club −9 and Broadcast −23 LUFS targets.
- Mastering
- Mastering is the final step before release: balancing a finished mix's tone, width and loudness so it sounds consistent on every speaker and service.
- Mastering One
- Mixing
- Mixing balances the tracks of a song (their levels, panning, EQ, dynamics and effects) into one stereo file.
- Mix One
- Panning
- Panning places a sound between the left and right speakers, giving each part its own space in a stereo mix.
- Reverb
- Reverb is the sound of a space: the reflections that make a voice sound as if it is in a room, a hall or a cathedral. A little adds depth; a lot pushes a sound into the background.
- Sidechainalso Sidechain compression, Ducking
- Sidechaining lets one sound control an effect on another, most often a kick drum ducking the bass or pads so the kick punches through: the pumping heard in dance music. Soundverse AI Studio's Sidechain One makes that pump on the tempo grid rather than from another track.
- True peakalso dBTP
- True peak is the highest level a signal reaches between its digital samples once it is played back. Keeping it a little under 0 dBTP, often at −1 dBTP, avoids distortion after streaming services convert the file.
AI music
- AI music generator
- An AI music generator creates original music from a description, a set of style tags or a reference track. On Soundverse, AI Music Generator makes instrumentals and AI Song Generator makes songs with vocals.
- AI Music GeneratorAI Song Generator
- Text to music
- Text to music means generating audio from a written prompt that describes the genre, mood, instruments and tempo you want.
- AI Music Generator
- Prompt
- A prompt is the written instruction you give an AI model. A specific prompt (genre, mood, instruments, tempo, vocal style) gets closer to what you hear in your head than a vague one.
- Prompt Bible
- Reference track
- A reference track is an existing song you point to as a model for a sound or style. Soundverse's Similar Song and Similar Music Generators make new music in the style of a reference you upload.
- Similar Song GeneratorSimilar Music Generator
- Inpainting
- Inpainting regenerates only a section of a song you select, such as one line or the second chorus, with new lyrics or a new style, and keeps the rest as it was. It saves remaking a whole track to fix one part.
- Inpainting
- Extensionalso Extend, Song extension
- Extending continues a track with newly generated music: a longer instrumental, or another verse with new lyrics. Soundverse's Extend Music returns the full track, the original plus the new part.
- Extend Music
- Stem separationalso Vocal remover, Source separation
- Stem separation uses AI to split a finished mix into its parts. Soundverse's Stem Separator offers 2 stems (vocals and backing), 6 (vocals, drums, bass, guitar, piano and other) or a maximum split with a MIDI file for each part.
- Stem Separator
- Voice clone
- A voice clone is an AI model of one singing voice made from a short recording. On Soundverse, Train Voice Clone needs at least 15 seconds of singing, and AI Song Generator can then sing new songs in that voice. Only clone a voice you own or have permission to use.
- Train Voice Clone
- Voice model
- A voice model is a more detailed AI model of a voice, trained on longer recordings. Soundverse's Train Voice Model uses at least 10 minutes of singing or speaking, and the finished model works in Voice Swap.
- Train Voice Model
- Voice swap
- Voice swap re-sings an existing vocal in a different voice model while keeping the original performance: the melody, timing and phrasing.
- Voice Swap
- Audio fingerprint
- An audio fingerprint is a compact signature of a recording that identifies it even after compression or trimming. Soundverse's Copyright Checker uses fingerprints to tell you whether a track matches a released recording.
- Copyright Checker
- Watermarkalso Audio watermark
- An audio watermark is a mark hidden in the sound itself, inaudible to listeners, that identifies where a file came from. Soundverse adds one to downloaded tracks.
Soundverse
- Agent One
- Agent One is Soundverse's AI music agent. You describe a goal in plain language and it plans the steps, picks the tools and runs them, from lyrics and songs to stems, cover art and music videos.
- Agent One
- AI Studio
- AI Studio is Soundverse's AI music studio: a multitrack timeline in the browser for audio, MIDI, video and image tracks, with AI generation on the timeline, eight built-in instruments, 28 effects and real-time collaboration.
- AI StudioWhat is an AI studio?
- Artist DNAalso DNA
- An Artist DNA is a Soundverse style model trained on a creator's own songs. You can generate songs, instrumentals or vocals from a DNA, and artists can train their own and earn when others export tracks made with it.
- DNATrain Artist Model
- DJ One
- DJ One is Soundverse's two-deck DJ booth in the browser. It plays the songs in your Soundverse Library, with sync, hot cues, loops, a 3-band EQ, performance effects and Automix.
- DJ One
- Trace
- Trace is Soundverse's music similarity search. It compares songs layer by layer, from the full mix and individual stems to sections, melody, lyrics and style, so it can surface a lifted melody or a reused vocal recording that fingerprinting alone would miss.
- Trace
- Mix One
- Mix One is Soundverse AI Studio's mixing stage. It places each track on a frequency and stereo map using one of 18 genre templates, and shows where tracks mask each other.
- AI Studio
- Mastering One
- Mastering One is Soundverse AI Studio's mastering chain on the master track. Press Master while the song plays and it listens to two seconds of the mix, then sets itself toward the genre and loudness target you pick.
- AI Studio
- Soundverse v9also Soundverse v8, Soundverse v7, Soundverse v6
- Soundverse's numbered model versions for songs and instrumentals are Soundverse v9, Soundverse v8, Soundverse v7 and Soundverse v6. Soundverse v9 is the newest, on the Pro plan and above; Soundverse v6 is on every plan, Free included.
- Soundverse v9Soundverse v6
Put the words to work
Hear the terms in action: make a song, then edit, mix and master it in AI Studio. Soundverse is free to start, with 1,000 tokens on sign-up.