Suno AI Music Generation: The Definitive Technical Reference

Updated March 3, 2026

TL;DR: Suno generates full songs — vocals, instruments, arrangement, and mix — from text prompts. V5 produces broadcast-quality audio at 48kHz with up to 4 minutes per generation. Master three systems (prompt text + metatags + Creative Sliders) and Suno becomes a production tool, not a novelty. Use Custom Mode for control, metatags for song structure, and the Song Editor for iterative refinement. Pro tier ($10/month) unlocks V5 and commercial rights. Credits don’t roll over.

Suno is the first AI music generation platform where the output is genuinely usable in production contexts. Not as a curiosity, not as a placeholder — as actual music that listeners engage with without realizing it was AI-generated. V5, released September 2025, crossed that threshold.¹

The difference between “interesting AI music” and “music I’d actually release” comes down to understanding three control systems:

1. Prompt text: Genre, mood, instrumentation, and vocal style described in natural language
2. Metatags: Structural directives like [Verse], [Chorus], [Bridge] that control arrangement
3. Creative Sliders: Weirdness, Style Influence, and Audio Influence that shape the generation’s personality

Most users type a sentence and click generate. The results are hit-or-miss because Suno’s default behavior optimizes for broad appeal, not for what you specifically want. This guide teaches you to be specific.

I’ve generated thousands of tracks across every genre Suno supports, tested every metatag combination documented and undocumented, and mapped the boundaries of what each model version handles well and poorly. This guide distills that experience into the definitive technical reference.

Table of Contents

Part 1: Foundations

1. What is Suno?
2. Getting Started
3. Models and Versions
4. Pricing and Credits

Part 2: Prompt Engineering

1. The Prompt Architecture
2. Genre and Style Descriptors
3. Vocal Styling
4. Instrumental Mode

Part 3: Song Structure

1. Metatags Reference
2. Structural Tags
3. Instrumental and Vocal Tags
4. Advanced Metatag Patterns

Part 4: Creative Controls

1. Creative Sliders
2. Song Editor
3. Covers and Remixes
4. Persona Voices

Part 5: Production Workflows

1. The Generation Loop
2. Suno Studio DAW
3. Stem Separation and Export
4. DAW Integration

Part 6: Advanced Techniques

1. Genre Blending
2. Multi-Section Composition
3. Prompt Chaining
4. Troubleshooting

Part 7: Business and Legal

1. Commercial Licensing
2. Copyright and Legal Landscape
3. Competitors and Alternatives

Part 8: Reference

1. API and Integration Status
2. Quick Reference Card
3. Changelog
4. References

What is Suno?

Suno is a generative AI platform that creates complete songs from text descriptions. Unlike DAWs, sample libraries, or loop-based tools, Suno generates every element of a track simultaneously: melody, harmony, rhythm, instrumentation, vocals (with lyrics), arrangement, and mix. You describe what you want; Suno produces a finished song.

What makes Suno different:

What you can create:

• Full songs with vocals: Any genre, any language, original lyrics or AI-generated
• Instrumentals: Background music, scores, ambient tracks
• Genre experiments: Cross-genre fusions that would require multiple specialist musicians
• Variations: Generate dozens of takes on the same concept, pick the best
• Production elements: Stems for use in traditional DAW workflows

What Suno is not:

• Not a DAW: You don’t mix, master, or arrange manually (though Studio adds some of this)
• Not deterministic: The same prompt produces different results each time
• Not a sample library: You can’t isolate and reuse individual sounds precisely
• Not unlimited: Generation costs credits, and quality varies between attempts

Getting Started

Quick start (5 minutes)

1. Create an account at suno.com. Free tier gives 50 credits per day (approximately 10 generations).

2. Try Simple Mode first. Type a short description like “upbeat indie rock song about a road trip” and click Create. Suno generates lyrics, melody, arrangement, and vocals automatically.

3. Switch to Custom Mode for control. Custom Mode separates the prompt into distinct fields:

4. Style of Music: Genre, mood, instrumentation descriptors
5. Lyrics: Your lyrics with metatags for structure

6. Title: Song title

7. Listen to both outputs. Suno generates two variations per creation. Pick the one closer to your intent, then refine.

8. Use Extend to continue a song beyond its initial generation, or Song Editor to replace specific sections.

Interface overview

Suno’s web interface has two primary creation modes:

Simple Mode: One text box. Describe the song in natural language. Suno infers genre, writes lyrics, and generates everything. Good for exploration, bad for precision.

Custom Mode: Three separate fields (Style, Lyrics, Title) plus Creative Sliders. This is where serious work happens. The Style field accepts genre and production descriptors. The Lyrics field accepts text with metatags. The sliders control generation personality.

Start with Custom Mode. Simple Mode is convenient but strips away the controls that make Suno useful for production work. Every technique in this guide assumes Custom Mode.

Models and Versions

Suno has iterated rapidly since launch. Each version brings meaningful quality improvements, but access varies by subscription tier.

Version timeline

Current model access

V5 is a meaningful upgrade over V4.5. The difference is most audible in vocal naturalness (less “AI singer” quality), low-frequency clarity (bass and kick separation), and stereo imaging. If you’re evaluating Suno for production use, evaluate on V5, not the free tier.

What V5 changed

V5 (internally called “chirp-crow”) represents Suno’s largest single-version improvement:¹

• 48kHz sample rate: Up from 44.1kHz. Audibly better high-frequency detail on studio monitors and quality headphones.
• Vocal naturalness: Reduced the “uncanny valley” quality that marked previous versions. Vibrato, breath sounds, and consonant articulation are more convincing.
• Instrument separation: Individual instruments in the mix are more distinct. Less “wall of sound” blending.
• Dynamic range: Better handling of quiet-to-loud transitions. Previous versions tended to compress everything.
• Genre accuracy: Better adherence to genre conventions. A “jazz” prompt sounds more authentically jazz, not “pop with jazz chords.”
• Suno Studio: In-browser DAW for post-generation editing. Mix adjustment, stem isolation, and arrangement changes without re-generating.²

Pricing and Credits

Verified as of March 2026. Suno pricing changes without notice. Check suno.com/pricing for current rates.

Plan comparison

Credit economics

Each generation costs approximately 5 credits and produces 2 song variations. A Pro subscription’s 2,500 monthly credits yields roughly 500 generations (1,000 song variations).

Credit-efficient practices: - Use Custom Mode with specific prompts to reduce throw-away generations - Extend promising tracks rather than re-generating from scratch - Use the Song Editor to fix sections rather than regenerating entire songs - Save credits by refining your Style prompt before generating

Credits do not roll over. Unused credits at the end of a billing cycle are lost. Plan your generation sessions accordingly.

The Prompt Architecture

Suno’s Custom Mode splits your creative input into three fields, each serving a distinct purpose. Understanding what goes where — and what doesn’t — is the difference between hit-or-miss results and consistent output.

The Style field

The Style field defines the musical character of your generation. It accepts natural language descriptors for genre, mood, tempo, instrumentation, vocal quality, and production style.

The optimal formula:

[Genre] [Subgenre], [Tempo/Energy], [Key instruments], [Vocal style], [Production quality], [Mood]

Example:

Indie folk rock, mid-tempo, acoustic guitar and mandolin, warm female vocals, lo-fi production, nostalgic and wistful

Descriptor sweet spot: 4–7 descriptors. Fewer than 4 gives Suno too much latitude. More than 7 and descriptors start competing with each other, producing muddy results where no single quality comes through clearly.³

What works in the Style field

What doesn’t work in the Style field

• Specific artist names: “Sounds like Adele” is unreliable and may be filtered. Use descriptive equivalents: “powerful female vocal, piano-driven pop ballad”
• Technical mixing terms: “Sidechain compression on the kick” — Suno doesn’t interpret mixing parameters
• Exact BPM control: “127 BPM” is treated as approximate guidance, not a metronome lock
• Negative instructions: “No drums” in the Style field is unreliable. Use the Instrumental toggle or metatags for structural control

The Lyrics field

The Lyrics field accepts your song text with optional metatags for structural control. Without metatags, Suno infers structure from line breaks and content patterns.

Basic lyrics (no metatags):

Walking down the empty road
Headlights fading in the rain
Every mile feels like a year
But I keep driving through the pain

Lyrics with metatags (recommended):

[Verse 1]
Walking down the empty road
Headlights fading in the rain

[Chorus]
Keep driving, keep driving
Through the storm and through the night

[Verse 2]
Every mile feels like a year
But the horizon's getting bright

[Chorus]
Keep driving, keep driving
Through the storm and through the night

[Outro]
And the sun comes up again

Always use metatags. Without them, Suno makes structural decisions that may not match your intent. A [Chorus] tag ensures repetition and melodic emphasis. A [Bridge] tag signals a harmonic departure. These structural cues dramatically improve output consistency.

The Title field

The Title field names your generation. It has minimal effect on the musical output but appears in metadata and Suno’s library. Keep it descriptive for your own organization.

Genre and Style Descriptors

Suno recognizes hundreds of genre and style terms. The model’s training data skews toward Western popular music, so genre accuracy varies by specificity and cultural origin.

High-confidence genres (consistent results)

These genres produce reliably accurate outputs because they’re well-represented in training data:

Medium-confidence genres (usable with guidance)

Low-confidence genres (requires iteration)

Vocal Styling

Vocal character is one of the most controllable aspects of Suno output. V5 significantly improved vocal naturalness and expressiveness.

Vocal descriptors

Combining vocal descriptors

Stack 2–3 vocal descriptors for precise control:

Raspy male vocals with subtle vibrato, lo-fi warmth

Ethereal female vocals, breathy and reverb-heavy, choir harmonies

Deep baritone, smooth jazz delivery, minimal processing

Language and multilingual vocals

Suno V5 supports multilingual vocal generation. The model infers language from your lyrics. For non-English lyrics:

• Write lyrics in the target language in the Lyrics field
• Optionally add the language to the Style field: “Japanese city pop, female vocals”
• Expect best results in English, Spanish, Portuguese, French, Japanese, Korean, and Mandarin
• Lesser-represented languages may produce accented or imprecise pronunciation

Instrumental Mode

Toggle Instrumental in Custom Mode to generate tracks without vocals. The Style field becomes the only creative input.

When to use instrumental mode

• Background music: Podcast intros, video scores, ambient work music
• Production elements: Beat beds, chord progressions, atmospheric textures
• Genre exploration: Test genre descriptors without vocal quality as a variable
• DAW integration: Generate backing tracks for live vocal recording

Instrumental prompt patterns

Without vocals, the Style field needs more descriptive detail to compensate:

Cinematic orchestral score, sweeping strings, French horns, timpani rolls, epic and triumphant, Hans Zimmer inspired

Lo-fi hip-hop beat, jazzy piano chords, vinyl crackle, mellow drums, study music

Ambient electronic, pad textures, slow evolving synths, ethereal and spacious, Brian Eno inspired

Tip: Even in instrumental mode, add [Instrumental] or [Instrumental Break] metatags in the Lyrics field to reinforce the intent and control arrangement structure.

Metatags Reference

Metatags are Suno’s structural control language. Placed in the Lyrics field inside square brackets, they direct arrangement, instrumentation, dynamics, and vocal behavior. This is where Suno transforms from a prompt-to-song toy into a composition tool.

How metatags work

Metatags are processed as arrangement directives, not as lyrics. When Suno encounters [Chorus], it: 1. Signals a section change in the arrangement 2. Applies typical chorus characteristics (melodic emphasis, fuller instrumentation, higher energy) 3. If the same [Chorus] text appears again, attempts to repeat the melody and arrangement

Metatags are case-insensitive: [VERSE], [Verse], and [verse] are equivalent.

Structural Tags

These tags define song sections and control arrangement flow.

Primary structural tags

Numbered sections

Use numbers to distinguish repeated section types:

[Verse 1]
First verse lyrics here

[Chorus]
Chorus lyrics

[Verse 2]
Second verse with different lyrics

[Chorus]
Same chorus lyrics (encourages melodic repetition)

Numbering verses helps Suno understand that each verse should have different melody while choruses should repeat their melody.

Instrumental and Vocal Tags

These tags control instrumentation and vocal behavior within sections.

Instrumental tags

Vocal tags

Advanced Metatag Patterns

Parameterized metatags

Metatags accept descriptive modifiers after a colon:

[Verse: whispered vocals, acoustic guitar only]
Walking through the morning mist
The world still sleeping, still

[Chorus: full band, powerful vocals]
But I'm awake, I'm alive
And every sound is a sign

The colon syntax lets you modify individual sections without changing the global Style field. This is the most powerful metatag feature — it gives you per-section control over arrangement.

Dynamic and production metatags

Combining structural and modifier tags

[Intro: ambient pads, reversed guitar, ethereal]
[Verse 1: lo-fi drums, muted bass, whispered vocals]
Words that float on morning air
Disappearing into light

[Pre-Chorus: building energy, adding layers]
But something shifts beneath the surface

[Chorus: full production, soaring vocals, epic drums]
We break through the silence
Into the wide open sky

[Bridge: stripped down, piano only, vulnerable vocals]
And in the quiet after the storm

[Outro: fade out, ambient reprise]

This pattern gives you DAW-level arrangement control through text alone.

Creative Sliders

Creative Sliders are V4.5+ controls that shape the generation’s personality. They appear in Custom Mode below the Lyrics field.

Weirdness

Range: Safe ← → Chaos (slider, no numerical values exposed)

When to increase Weirdness: - Experimental or avant-garde genres - When conventional results feel generic - For genre-blending experiments

When to decrease Weirdness: - Commercial music that needs to sound “normal” - When working within strict genre conventions - For background/ambient music that shouldn’t draw attention to itself

Style Influence

Range: Loose ← → Strong (slider)

Use Strong when your Style field is precise and you want exactly what you described. Use Loose when you want Suno to interpret your prompt more freely and potentially surprise you.

Audio Influence

Range: Controls how much any uploaded audio reference affects generation.

Available when using Audio Upload (Covers, Remixes, or Add Vocals/Instrumentals). Higher values make the output more closely follow the reference audio’s characteristics.

Song Editor

The Song Editor enables post-generation editing without re-creating the entire song. This is Suno’s answer to the “90% perfect but one section is wrong” problem.

Available operations

Inpainting workflow

1. Select the time range to replace (drag on the waveform)
2. Optionally provide new lyrics/metatags for the replacement section
3. Generate — Suno creates new content that matches the surrounding audio
4. Listen and compare. Accept or regenerate.

Inpainting is iterative. Rarely does the first replacement perfectly match the surrounding context. Budget 2–5 attempts for seamless results.

Extend workflow

1. Click Extend on any existing generation
2. Optionally provide lyrics/metatags for the continuation
3. Suno generates ~30–60 seconds of new audio that continues from the endpoint
4. Each extension is a separate generation (costs credits)

Best practice: Include a structural metatag at the start of your extension prompt (e.g., [Chorus] or [Outro]) to guide what the extension generates.

Covers and Remixes

Pro and Premier tiers can create covers and remixes of existing Suno tracks.

Covers

Upload or select an existing Suno track as a reference, then apply a new style:

Style: Acoustic folk cover, fingerpicked guitar, soft female vocals, intimate production

The cover maintains the melody and lyrics but reimagines the arrangement and production.

Remixes

Remixes take an existing track and transform it more aggressively than covers:

Style: EDM remix, heavy bass, 128 BPM, drop-focused, festival energy

Add Vocals / Add Instrumentals

Two specialized modes that layer onto existing audio:

• Add Vocals: Upload an instrumental track, Suno generates vocals over it
• Add Instrumentals: Upload a vocal track, Suno generates instrumentation behind it

These are powerful for integrating Suno into traditional production workflows — record real vocals and let Suno generate the backing track, or vice versa.

Persona Voices

Persona Voices (Pro/Premier) let you create and reuse consistent vocal characters across generations. Instead of hoping each generation assigns a similar voice, you define a persona and reference it.

Creating a Persona Voice

1. Generate a song with vocals you like
2. Select “Create Persona” from that generation
3. Name and save the persona
4. Reference it in future generations

Using Persona Voices

In Custom Mode, select your saved persona from the Persona dropdown. All subsequent generations in that session use that vocal character for consistency.

Limitations: - Persona Voices capture timbre and basic delivery style, not exact vocal technique - Results vary when applying a persona far outside its original genre - Persona Voices are account-specific — they can’t be shared

The Generation Loop

Effective Suno usage follows an iterative workflow, not a single-prompt approach.

The production cycle

1. IDEATION
   ↓ Generate 5-10 variations with different Style descriptors
   ↓ (Cost: ~25-50 credits)

2. SELECTION
   ↓ Pick the 1-2 best results
   ↓ Identify what works and what doesn't

3. REFINEMENT
   ↓ Adjust Style descriptors based on what you heard
   ↓ Refine lyrics and metatags
   ↓ Regenerate with tighter prompts
   ↓ (Cost: ~15-30 credits per round)

4. EXTENSION
   ↓ Extend the best track to full length
   ↓ Add missing sections (bridge, outro)
   ↓ (Cost: ~5-15 credits)

5. EDITING
   ↓ Use Song Editor to fix weak sections
   ↓ Inpaint, crop, fade as needed
   ↓ (Cost: ~5-20 credits)

6. EXPORT
   ↓ Download final audio (MP3/WAV)
   ↓ Optionally export stems for DAW work

Typical cost for a polished track: 50–100 credits (10–20 generations across all stages).

Credit-efficient workflow tips

1. Spend time on the prompt, not on generations. A well-crafted Style + Lyrics prompt produces better first results than rapid iteration with vague prompts.
2. Generate in batches. When exploring a concept, generate 4–6 variations at once, then pick the best direction before refining.
3. Use the Song Editor over regeneration. If 80% of a track is good, edit the remaining 20% rather than regenerating the entire song.
4. Save successful Style prompts. When a particular descriptor combination works well, save it for reuse.

Suno Studio DAW

Suno Studio (Premier tier, launched with V5) is an in-browser digital audio workstation for post-generation editing. It bridges the gap between Suno’s generation engine and traditional music production.²

Studio capabilities

Studio 1.2 (February 2026)

The latest Studio update added:⁴

• Warp markers: Micro-adjust timing of individual notes and phrases
• Remove FX: Strip AI-applied reverb and delay to get dry stems
• Alt takes: Generate and audition alternative sections inline
• Expanded time signature support: Better handling of 3/4, 6/8, and odd time signatures

When to use Studio vs. export to DAW

Stem Separation and Export

Suno offers two levels of stem separation:

2-stem separation (all tiers)

Separates audio into: - Vocals: All vocal content - Instrumental: Everything else

Useful for: karaoke versions, vocal sampling, basic remixing.

12-stem separation (Pro/Premier)

Separates audio into up to 12 individual stems: - Vocals, drums, bass, guitar, keys/piano, synths, strings, brass, woodwinds, percussion, effects, other

Quality notes: AI-based stem separation is imperfect. Expect bleed between stems, especially between similar-frequency instruments. The separation quality improved significantly in V5 but still doesn’t match purpose-built tools like iZotope RX or Demucs on clean source material.

Export formats

• MP3: Standard compressed audio. Good for sharing, streaming, and drafts.
• WAV: Uncompressed audio. Required for professional DAW work and mastering.

DAW Integration

Suno’s output integrates into traditional production workflows via stem export.

Recommended workflow

1. Generate in Suno until the arrangement and vibe are right
2. Export 12 stems (Pro/Premier) as WAV files
3. Import into your DAW (Logic Pro, Ableton, Pro Tools, FL Studio, Reaper)
4. Mix and master with professional tools and processing
5. Replace or augment individual stems with live recordings if needed

What you gain from DAW mixing

• EQ and compression: Per-stem tonal shaping that Suno’s AI mixing doesn’t provide
• Spatial processing: Precise stereo placement, reverb sends, delay throws
• Automation: Dynamic changes over time (fade builds, filter sweeps)
• External instruments: Layer live recordings with AI-generated stems
• Mastering chain: Loudness normalization, limiting, final polish for release

Genre Blending

One of Suno’s unique strengths is generating music at genre intersections that would require multiple specialist musicians in traditional production.

Effective blending patterns

Two-genre fusion (most reliable):

Jazz-funk fusion, slap bass, Rhodes piano, syncopated drums, groovy and sophisticated

Genre + era mashup:

80s synthwave meets modern trap, analog synths, 808 bass, retro-futuristic

Genre + unexpected instrument:

Death metal with jazz saxophone solos, blast beats, dissonant chords

Blending rules

1. Lead with the dominant genre. “Jazz with electronic elements” produces different results than “Electronic with jazz elements.”
2. Limit to 2–3 genres. More than that and Suno’s output becomes an unfocused compromise.
3. Use era markers to anchor style. “90s” or “2020s” helps Suno pick the right production conventions.
4. Increase Weirdness for unusual fusions. The default Weirdness setting tries to normalize everything, which defeats the purpose of genre blending.

Multi-Section Composition

For songs longer than 4 minutes, you need to compose in multiple generations and join them.

Strategy 1: Extend

Generate the first section, then use Extend to add subsequent sections. Each extension uses the ending of the previous section as context.

Pros: Musical continuity. Each extension naturally follows the previous. Cons: Less control over later sections. Musical drift over multiple extensions.

Strategy 2: Section-by-section generation

Generate each section independently with specific metatag + Style combinations, then join in a DAW.

Pros: Maximum control over each section’s character. Cons: Transitions between independently generated sections can sound jarring. Requires DAW skills for joining.

Strategy 3: Hybrid approach (recommended)

1. Generate the core of the song (verse-chorus-verse-chorus) as one generation
2. Extend for the bridge and final chorus
3. Use Song Editor to inpaint any weak transitions
4. Export stems and finalize in a DAW

Prompt Chaining

Build complex songs through a sequence of related generations.

Chain pattern

Generation 1: "Atmospheric intro, ambient pads, slow build"
   → Extend with: "[Build-Up] [Drop: full energy, heavy drums]"
   → Extend with: "[Verse 1: vocals enter, riding the beat]"
   → Extend with: "[Chorus: anthemic, crowd-singing energy]"
   → Extend with: "[Outro: fade out, return to ambient pads]"

Each extension inherits the musical DNA of the previous generation, creating a coherent multi-section composition without starting from scratch each time.

Troubleshooting

Common problems and solutions

Generation quality checklist

Before spending credits on refinement, verify your prompt covers:

• [ ] Genre is specific (not just “rock” but “indie rock” or “post-punk”)
• [ ] Vocal style is described (or Instrumental is toggled on)
• [ ] Metatags define structure (at minimum: Verse, Chorus, Outro)
• [ ] 4–7 descriptors in the Style field (not too few, not too many)
• [ ] Mood is explicit (Suno defaults to upbeat/positive without guidance)

Commercial Licensing

Verified as of March 2026. Licensing terms change. Check Suno’s current Terms of Service for binding language.⁵

What each tier allows

Important caveats

Copyright protection for 100% AI content is legally unsettled. As of 2026, purely AI-generated music may not qualify for copyright registration in the US. This means: - You have commercial rights (Suno grants you a license) - But you may not be able to prevent others from using the same or similar output - Adding human creative elements (original lyrics, live instrument recordings, arrangement choices in a DAW) strengthens your copyright claim

Revenue is yours. Pro and Premier users keep 100% of revenue from Suno-generated music. Suno does not claim royalties or revenue share.⁵

Copyright and Legal Landscape

AI music generation exists in an evolving legal environment.

Key legal developments

• Warner Music settlement (November 2025): Warner settled its lawsuit against Suno. Terms undisclosed but Suno continues operating.⁶
• UMG and Sony lawsuits: Major label lawsuits against Suno remain active as of early 2026. Claims center on alleged use of copyrighted recordings in training data.⁶
• Udio/UMG settlement (2025): Competitor Udio settled with UMG, establishing some precedent for the industry.⁷
• US Copyright Office: Has stated that purely AI-generated works cannot be copyrighted, though works with sufficient human authorship containing AI elements may qualify.⁸

Practical guidance

1. Don’t use Suno to replicate specific copyrighted songs. The Covers feature is designed for covering Suno-generated tracks, not commercial recordings.
2. Add human creative elements to strengthen copyright claims: write original lyrics, record live instruments over Suno stems, make arrangement decisions in a DAW.
3. Document your creative process. If your work is ever challenged, evidence of human creative choices strengthens your position.
4. Stay current on legal developments. This area is changing rapidly.

Competitors and Alternatives

When to use what

• Full songs with vocals: Suno (V5)
• Stems for DAW production: Udio (highest stem quality)
• API-driven generation: Stable Audio (only platform with a public API)
• Sound design and SFX: Stable Audio
• Film scoring: AIVA (MIDI export for orchestral editing)

API and Integration Status

Verified as of March 2026.

Suno does not offer a public API. There is no official REST API, SDK, or programmatic access for individual users or developers.

What exists

For developers

If you need programmatic music generation: - Stable Audio: Has an official API with documented endpoints - Replicate: Hosts open-source music generation models with API access - Custom deployment: Open-source models like MusicGen (Meta) can be self-hosted

Quick Reference Card

Custom Mode template

STYLE FIELD:
[Genre] [Subgenre], [Tempo], [Key instruments], [Vocal style], [Production], [Mood]

LYRICS FIELD:
[Intro: descriptors]

[Verse 1]
Your lyrics here

[Pre-Chorus]
Building lyrics

[Chorus]
Hook lyrics

[Verse 2]
More lyrics

[Chorus]
Same hook (for melodic repetition)

[Bridge: contrasting descriptors]
Different energy lyrics

[Chorus]
Final hook

[Outro: fade out]

Essential metatags

Creative Sliders cheat sheet

Pricing quick reference

Changelog

References