AI music creation tools

WhyLabel: Establishing access to the industry-leading generative AI music platforms is the first step to creation.

HowLabel:

• Sign up for accounts on both platforms to compare their distinct sonic signatures.
• Explore the 'Custom Mode' in Suno and 'Advanced Features' in Udio to move beyond simple random generation.
• Familiarize yourself with the credit systems to manage your daily generation limits.

DoneWhenLabel: Accounts are active and the user interface for 'Custom' generation is open.

WhyLabel: Precise prompting prevents generic-sounding results and ensures the AI follows your creative vision.

HowLabel:

• Use specific genre tags (e.g., '90s East Coast Hip Hop' instead of just 'Rap').
• Include technical descriptors like 'BPM', 'Key' (e.g., 'C Minor'), and 'Production Style' (e.g., 'Lo-fi', 'High Fidelity', 'Reverb-heavy').
• Use structural tags like [Intro], [Verse], [Chorus], [Bridge], and [Outro] to guide the AI's composition flow.

DoneWhenLabel: You have generated three 30-second clips that accurately match three different complex style prompts.

WhyLabel: AI tools usually generate music in short bursts; extending is the only way to build a full-length song.

HowLabel:

• Select a 30-second clip you like and use the 'Extend' feature.
• Set the 'Extend From' timestamp to a point where a transition should occur (e.g., the end of a chorus).
• Change the prompt for the extension to introduce new elements like a 'Guitar Solo' or 'Drum Breakdown'.

DoneWhenLabel: A single cohesive track of at least 2 minutes created through multiple extensions.

WhyLabel: Consistency in technical practice builds an intuitive feel for how the AI responds to specific keywords.

HowLabel:

• Spend 15 minutes daily testing 'Negative Prompts' (what you don't want, e.g., 'no vocals', 'no drums').
• Spend 15 minutes daily attempting to recreate a specific reference song's vibe using only text.
• Document which keywords yield the most 'human-like' results.

DoneWhenLabel: Completion of 7 consecutive days of documented prompting practice.

WhyLabel: AI-generated lyrics can be repetitive; providing your own ensures originality and emotional depth.

HowLabel:

• Follow a standard pop structure: Verse 1, Chorus, Verse 2, Chorus, Bridge, Chorus, Outro.
• Use rhythmic patterns and rhymes that are easy for the AI to 'sing' (avoid overly complex tongue-twisters).
• Insert [Brackets] for vocal instructions like [Whispered] or [Belting].

DoneWhenLabel: A complete lyric sheet for one full song is ready in a text document.

WhyLabel: The first generation sets the DNA (tempo, key, vocal timbre) for the entire project.

HowLabel:

• Input your Verse 1 and Chorus lyrics into the 'Custom' box.
• Apply your refined style prompt from Phase 1.
• Generate multiple versions (at least 10) until you find the perfect vocal character and melody.

DoneWhenLabel: One 'Seed' clip (approx. 60s) that serves as the foundation for the song.

WhyLabel: Creating a small body of work allows you to practice different genres and production styles.

HowLabel:

• Song 1: High-energy electronic or pop track.
• Song 2: Acoustic or 'unplugged' ballad to test vocal realism.
• Song 3: Experimental or genre-bending track (e.g., 'Jazz-Metal').
• Ensure each song is at least 3 minutes long.

DoneWhenLabel: Three fully realized, high-quality audio files (WAV or MP3).

WhyLabel: AI outputs are usually a single stereo file; separating them into 'stems' (vocals, drums, bass) allows for professional mixing.

HowLabel:

• Use an open-source stem separation tool (e.g., Ultimate Vocal Remover or a browser-based generic stem splitter).
• Export the separated tracks as high-quality WAV files.
• Check for 'artifacts' (digital noise) in the vocal stem.

DoneWhenLabel: A folder containing separate audio files for Vocals, Drums, Bass, and Other instruments.

WhyLabel: AI music often has 'muddy' frequencies; manual EQing brings clarity and professional punch.

HowLabel:

• Import stems into a free Digital Audio Workstation (DAW).
• Apply a 'High Pass Filter' to the vocals and instruments to remove low-end rumble.
• Use a compressor on the vocal track to make it sit consistently in the mix.
• Add a touch of 'Plate Reverb' to the vocals for a professional finish.

DoneWhenLabel: A balanced mix where every instrument is clearly audible.

WhyLabel: Mastering ensures your track is loud enough for streaming platforms and sounds good on all speakers.

HowLabel:

• Use a generic 'Auto-Mastering' tool or a 'Limiter' plugin in your DAW.
• Target a loudness level of approximately -14 LUFS (the standard for most streaming services).
• Ensure there is no 'clipping' (distortion) in the final output.

DoneWhenLabel: A final, mastered WAV file ready for distribution.

WhyLabel: A fixed deadline creates accountability and allows for a structured marketing buildup.

HowLabel:

• Pick a Friday (standard release day) 4 weeks from today.
• Mark this as your 'Live Stream' or 'Social Media Launch' date.

DoneWhenLabel: A specific date is marked on your calendar.

WhyLabel: Music needs a visual identity to be noticed on digital platforms.

HowLabel:

• Use a generic AI image generator to create 'Cover Art' based on the song's mood.
• Create a 15-second 'Canvas' or 'Visualizer' video for social media using the song's hook.

DoneWhenLabel: One high-res cover image (3000x3000px) and one teaser video.

WhyLabel: Building anticipation increases the reach of your final release.

HowLabel:

• Post the 15-second teaser on social media.
• Share a 'Behind the Scenes' look at your prompting process or stem separation.
• Engage with music communities to gather early feedback.

DoneWhenLabel: At least three social media posts published across different platforms.

WhyLabel: A live event provides a focal point for your audience to experience the music together.

HowLabel:

• Schedule a 30-minute live stream on a video platform.
• Play your 3-song repertoire.
• Explain the 'Technique' behind each song (e.g., the prompts used).
• Collect real-time reactions and answer questions about AI music.

DoneWhenLabel: Completion of a live broadcast with at least one listener.