What defines this Toronto rap sound
The traits people mean when they say this sound are: muted, dampened synth pads rather than bright ones; a sparse, deep 808 bass that leaves gaps instead of filling every beat; hi-hats kept minimal or brushed rather than busy; and a vocal that shifts between rapping and singing a melody, often called a half-sung flow. The mix tends to sit dark and spacious, with reverb (an echo-like sense of room) used to create distance rather than closeness.
Which tempo and key to use
Most tracks in this style sit between 78 and 95 BPM, slower than boom bap's usual 85 to 95 range and closer to half-time trap. Minor keys dominate, especially G minor and D minor, which give the muted pads their moody colour. Pick one tempo and one key per prompt; changing both at once makes it harder to judge what moved the sound.
| Element | Typical range |
|---|---|
| Tempo | 78-95 BPM |
| Key | G minor, D minor, C minor |
| 808 pattern | sparse, long sustained notes |
| Hi-hats | minimal or brushed, rarely fast rolls |
How to describe the vocal without naming anyone
Say what the voice does, not who it sounds like: half-sung half-rapped, breathy, melodic, low-register male vocal. In our measurements on ACE-Step 1.5, whisper-style vocal words reliably lowered loudness and belting reliably raised it, so a term like breathy or half-sung will move the render in a predictable direction rather than being ignored.
Avoid piling on more than two or three vocal descriptors at once, since a hierarchy of a few clear traits reads better than a long list of adjectives.
Common problems and how to fix the prompt
If the result comes out too bright or upbeat, add muted or dark to the pad description and drop any energetic mood words. If the 808 disappears or sounds thin, describe it as deep and sparse rather than distorted, since our word tests found some bass descriptors (distorted_bass, sub_bass) did not reliably change the low end on their own. If the vocal raps the whole way through with no melody, add half-sung or melodic explicitly, since that is a directional cue the model can act on.
Lyrics skeleton
Original, for structure
[Verse] Streetlights blur through the window glass Counting hours that never last [Chorus] Hold the silence close tonight Nothing left but city light [Bridge] Echo fading in the dark Still I feel it in my heart [Outro] Fading out, no words to say
Common mistakes
- Naming the artist directly, which some tools may ignore or handle unpredictably: describe muted pads, sparse 808, half-sung flow instead.
- Mixing contradictory moods like dark and upbeat in one prompt: pick one mood, such as moody or wistful, not both.
- Overloading hi-hats and percussion words when the style calls for minimal drums: keep drum descriptors to one or two, favouring space over density.
- Using quality words like best or masterpiece, which our tests treat as placebo: replace with specific sound descriptors instead.
- Forgetting to say the vocal type when writing a song prompt: always include male or female and a texture word like breathy or gravelly.
About the name
Drake is named here only because people search for this sound using his name. Lyro Music has no connection to Drake, his label or his estate; the prompts on this page describe general production traits of a Toronto rap era rather than copying any voice.