The audio generation revolution is here. With platforms like Suno and Udio for music creation, and ElevenLabs for realistic voice cloning, prompt engineering has expanded into acoustics. Generating a hit song or a natural-sounding voiceover requires mastering genre, mood, and structural tags.
1. Prompting for AI Music (Suno & Udio)
When prompting for music generators, you describe the style of the song rather than the lyrics. Use a combination of:
- Genres & Subgenres: E.g.,
"synthwave, 1980s retro, dream pop". - Instruments: E.g.,
"analog synthesizers, clean electric guitar, punchy drum machine". - Tempo & Mood: E.g.,
"120 BPM, melancholic, driving, energetic".
2. Structuring Lyrics and Song Sections
Both Suno and Udio support structural meta-tags in their lyric boxes. Place these in brackets to guide the arrangement of your song:
[Verse]
[Chorus]
[Guitar Solo]
[Bridge]
[Outro]This tells the AI when to build energy, when to introduce vocals, and when to play instrumentals.
3. Crafting Voiceovers and Audio Effects (ElevenLabs)
ElevenLabs produces lifelike text-to-speech. To get the best emotive performance, use formatting tricks in your text input:
- Punctuation is key: Use ellipses (
...) or dashes (-) to force natural pauses and breathing space. - Capitalization for emphasis: Writing words in CAPITAL letters can sometimes guide the AI to speak with more intensity or volume.
- Stability settings: Lower stability for more emotional range, and higher stability for consistent, calm voiceovers.