Why not just name the artist in the prompt
Putting an artist name in a prompt box can pull a model toward a style, but it also makes the output unpredictable and hard to repeat, and it is not something Lyro Music prompts do. Naming the production traits instead, gated drums, analogue pads, falsetto vocal, gives a result you can describe, adjust and reuse without depending on a name the model may or may not respect.
What is a 'gated drum' sound and why does it matter here
A gated drum sound is a snare or drum hit that gets cut off abruptly by a gate effect (a tool that silences audio once it drops below a set level), instead of ringing out naturally. It is the single most recognisable trait of this dark 80s synth-R&B sound, so naming it directly, as 'gated snare' or 'gated drums', does more work in a prompt than any mood word.
How does this compare to a synthwave prompt
This sound sits close to synthwave (see our synthwave prompt guide), which runs 80 to 120 BPM with darker, driving styles up to 130 BPM in minor keys such as A minor, D minor or E minor. The R&B version pulls the tempo down towards the slower end, adds a falsetto-leaning vocal and leans on sub bass and pads rather than arpeggios as the main hook.
| Trait | Synthwave | This sound |
|---|---|---|
| Tempo | 80-130 BPM | 80-120 BPM, often ~100 |
| Lead element | arpeggio, lead synth | vocal, pad swells |
| Drums | driving, electronic | gated, snappier snare |
| Mood | retro, cinematic | nocturnal, brooding, intimate |
How do I describe the vocal without naming a singer
Use timbre words: falsetto (singing in a high, light register above the natural chest voice), breathy, raspy, or moody male vocal. Adding 'falsetto-leaning' rather than 'falsetto' throughout gives a model room to move between registers, which tends to sound more natural over a full song than a vocal locked in falsetto the whole way through.
Can Lyro Music make this kind of track
Yes. Describe the sound, or start from a vocal idea you record, and Lyro Music turns it into a mixed and mastered song between 30 seconds and 3 minutes. Sung lyrics, including Turkish, route to Google's Lyria 3 Pro; vocal chops, tempo-locked beats and audio-input work route to ACE-Step 1.5. Try the free prompt generator at /ai-music-prompt-generator to build a prompt like the ones above.
Lyrics skeleton
Original, for structure
[Verse] Streetlights blur through the window glass Midnight calls but I let it pass [Chorus] Hold me close in the neon haze Lost in you till the end of days [Bridge] Every shadow knows my name Still I'm chasing this same refrain
Common mistakes
- Naming the artist directly, which is unpredictable and unreusable: describe gated drums, pads and falsetto vocal instead.
- Piling on more than 7 descriptors, which muddies the result: pick one from each of drums, bass, pad, vocal and mood.
- Mixing moods like 'upbeat and melancholic', which the model struggles to reconcile: pick one dominant mood, such as moody or nocturnal.
- Forgetting the tempo, which lets the model default to a random BPM: state 80 to 120 BPM depending on how driving you want it.
- Adding percentages or dB numbers to push an effect harder: in our measurements these numbers are not read, so use plain words like 'deep' or 'gated' instead.
About the name
The Weeknd is referenced here only to identify a well-known dark 80s synth-R&B sound; Lyro Music has no connection to the artist, and the prompts above describe general production traits of that era rather than copying any voice.