How to Use Mureka AI to Make Songs: Complete Guide (2026)

Mureka AI is a full-stack music generation platform that goes beyond simple text-to-song tools. It covers lyrics writing, song generation, instrumental creation, stem separation, MIDI export, and even voice cloning. The name combines "Music" and "Eureka," and the platform has attracted nearly 10 million users since its 2024 launch.

This tutorial walks through every major feature with practical steps, real prompt examples, and tips for getting better results. Whether you are creating your first AI song or looking to use Mureka's stem separation in a DAW workflow, this guide covers exactly how to do it.

What You Need Before Starting

Mureka is available on web, iOS, and Android. You can sign up for a free trial with limited credits. The Basic plan at $8/month (annual billing) gives you approximately 400 songs per month with commercial licensing.

For this tutorial, the free trial credits are enough to follow along with every section.

Mode 1: Song Creation (Text-to-Song and Lyrics-to-Song)

Song creation is Mureka's core feature. You have two approaches: describe what you want in plain text, or write specific lyrics and let Mureka compose music around them. The lyrics-first workflow is where Mureka genuinely stands out.

Step 1: Write or Paste Your Lyrics

Mureka works best when you provide structured lyrics. Use standard structure tags to tell the AI how your song should be organized:

• [Intro] - Opening instrumental or vocal section
• [Verse] - Main storytelling sections
• [Chorus] - The repeated hook of your song
• [Bridge] - A contrasting section, usually before the final chorus
• [Outro] - Closing section

Here is an example of well-structured lyrics:

[Intro]

[Verse 1]
Walking through the city lights alone
Every street reminds me of your phone calls
Late at night when nothing felt like home
You were the voice that caught me when I'd fall

[Chorus]
But now the silence fills the room
And I keep reaching out for you
A ghost of what we used to be
Still echoing inside of me

[Verse 2]
Coffee shops and corners where we sat
Your laughter hanging in the morning air
I keep looking back like you'll be there
But time moves on and I know that

[Chorus]
But now the silence fills the room
And I keep reaching out for you
A ghost of what we used to be
Still echoing inside of me

[Bridge]
Maybe someday I'll stop searching
For your shadow in the crowd
Maybe someday this ache will quiet down

[Outro]

Step 2: Choose Your Style and Genre

Be specific with your style description. Vague descriptions like "pop song" give the AI too much freedom to wander. Specific descriptions produce more consistent results.

Vague (less consistent):

Pop song, emotional

Specific (better results):

Indie pop, acoustic guitar driven, female vocals, melancholic but hopeful, 95 BPM, soft reverb on vocals

Good style elements to specify:

• Genre and subgenre (indie pop, lo-fi hip-hop, progressive rock)
• Primary instruments (acoustic guitar, piano, synth pads)
• Vocal character (soft female, raspy male, choir backing)
• Mood and energy (melancholic, uplifting, aggressive, dreamy)
• Tempo if you have a preference (slow ballad, mid-tempo groove, uptempo dance)

Step 3: Select a Model

Mureka offers several models:

• V8 - Latest general-purpose model, best overall quality
• O2 - Newer model, good for experimentation
• V7.5 - Stable and reliable, good for consistent output
• V6 - Older model, less refined but sometimes produces interesting results

Start with V8 for your first generations. If you are not getting the sound you want, try the same lyrics and style on O2 or V7.5. Different models interpret the same input differently, and sometimes an older model nails a particular genre better.

Step 4: Generate and Review

Hit generate and wait for the output. Mureka uses MusiCoT (Music Chain-of-Thought) technology, which means it plans the song structure before generating audio. This typically produces more coherent arrangements than models that generate audio sequentially.

Listen to the full track. Pay attention to:

• Do the musical phrases align with your verse and chorus breaks?
• Does the mood match your style description?
• Are the vocals clear and well-mixed?

If the first generation is not right, generate again with the same inputs. AI generation has inherent randomness, and the second or third attempt may be significantly better. You can also tweak your style description based on what the first output got wrong.

Mode 2: Melody Idea (Humming to Song)

This is Mureka's most unique feature. If you have a melody in your head but cannot write sheet music or play an instrument, you can hum it directly and let Mureka build a complete song around your melody.

Step 1: Record or Upload Your Melody

You can either record directly in the app or upload an audio file. Your recording does not need to be polished. Hum, whistle, or sing a rough melody. Even 10-15 seconds is enough for Mureka to work with.

Step 2: Add Style Direction

After uploading your melody, describe the style you want the full song to have. The AI will preserve your melodic contour while building instrumentation, harmony, and production around it.

Example style prompt:

Dreamy synth-pop, ethereal female vocals, reverb-heavy, 80s inspired, atmospheric

Step 3: Generate and Iterate

The generated track should follow the general shape of your melody while adding professional arrangement. If the AI drifted too far from your original idea, try a shorter melody clip or a more specific style description.

Mode 3: Instrumental Generation

Mureka generates instrumentals without vocals, useful for background music, beats, or production templates.

Step 1: Describe Your Instrumental

Skip the lyrics entirely and provide a detailed style description:

Cinematic orchestral piece, building tension, strings and brass,
film trailer style, dark and epic, timpani hits on transitions

Or for something more contemporary:

Lo-fi hip-hop beat, dusty vinyl crackle, mellow jazz piano chords,
head-nodding groove, 85 BPM, late night study vibes

Step 2: Select Model and Generate

The same model recommendations apply. V8 typically handles instrumentals well. For cinematic and orchestral pieces, try multiple models since each interprets orchestral textures differently.

Mode 4: Extend

Extend lets you continue a previously generated track. If you have a 2-minute song and want to add another verse or an extended outro, Extend builds on the existing audio rather than generating from scratch.

Step 1: Select a Generated Track

Choose any track you have previously generated in Mureka.

Step 2: Provide Direction for the Extension

Describe what the extension should contain:

Add a final chorus with building energy, then fade out with
just acoustic guitar and soft vocals

You can also paste additional lyrics for the extended section using the same structure tags.

Step 3: Generate the Extension

The AI will match the style, tempo, and instrumentation of the original track and continue from where it ended. This is how you build tracks longer than what a single generation produces, up to 5 minutes total.

Mode 5: Stem Separation (Audio Tools)

Stem separation splits any song into individual component tracks. This works on any uploaded audio, not just Mureka-generated tracks.

Step 1: Upload Your Audio File

Upload the song you want to separate. Most common audio formats are supported.

Step 2: Choose Separation Options

Mureka separates audio into four stems:

• Vocals - Isolated singing and vocal effects
• Drums - Kick, snare, hi-hats, and percussion
• Bass - Bass guitar, sub-bass, and low-end elements
• Other - Everything else (guitars, keys, synths, strings)

Step 3: Download Individual Stems

After processing, download each stem as a separate audio file. You can use these in your DAW for remixing, sampling, or creating karaoke versions.

Pro tip: Combine stem separation with MIDI export. Separate a track into stems, export the MIDI data, and import both into your DAW. This gives you the actual audio stems plus an editable MIDI representation that you can modify, re-instrument, or use as a compositional reference.

Mode 6: Style Matching

Style matching lets you upload a reference track and generate new music with a similar sound.

Step 1: Upload a Reference Track

Choose a song that represents the style you want. The AI analyzes the genre, tempo, instrumentation, and production style.

Step 2: Add Your Creative Direction

Provide lyrics or describe what you want the new track to be about. The style will come from the reference; the content comes from your input.

[Verse]
Neon signs reflecting on the wet pavement
Another Friday night without a destination
The city hums a melody I almost recognize
A familiar stranger underneath the station lights

[Chorus]
Take me where the music never stops
Where every beat is something that I lost
Dancing with the memory of sound
Spinning round and round and round

Step 3: Generate

The output will share stylistic DNA with your reference track while being an original composition based on your lyrics and direction. This is useful for creating cohesive albums or matching a specific sonic aesthetic without writing elaborate style descriptions.

Tips for Better Results

Be specific with style descriptions. "Rock song" gives the AI almost nothing to work with. "Garage rock, raw distorted guitars, shouted vocals, fast tempo, Stooges-inspired energy" gives it a clear target.

Use structure tags consistently. Always include [Verse], [Chorus], and other tags in your lyrics. Mureka's MusiCoT technology uses these to plan the musical arrangement.

Try multiple models. Generate the same lyrics and style on V8, O2, and V7.5. Each model has different strengths, and the best output varies by genre and style.

Keep first-generation sessions focused. Quality can drop after multiple consecutive generations. If you notice a dip, take a break or start a new session.

Use Melody Idea for specific melodies. If you have a tune in your head, humming it will always produce a closer result than trying to describe a melody in text. Even a rough recording works.

Export stems and MIDI for DAW work. Mureka is most powerful when used as a starting point rather than a finished-product generator. Generate, export, and refine in your preferred DAW.

Iterate on style, not just lyrics. If a generation sounds wrong, the problem is usually the style description, not the lyrics. Adjust your genre, instrument, and mood keywords first.

Using Mureka Through Musci.io

Musci.io provides access to Mureka alongside other generators including Suno, Udio, ElevenLabs, Minimax, ACE-Step, and Lyria 3. Through Musci.io, all of Mureka's key modes are available: Song, Instrumental, Extend, Stem Separation, Speech, and Podcast.

The unified interface lets you try the same concept across multiple AI music generators and compare results. You might find that Mureka produces the best stems for your DAW work while Suno gives you a more polished vocal track from the same lyrics.

FAQ

What is the best Mureka model to start with?

Start with V8 for the best overall quality. If you are not satisfied with the output, try the same input on O2 for a different interpretation. V7.5 is the most stable and predictable if you want consistent results. V6 is older and less refined but can occasionally produce interesting outputs for specific genres.

How long can a Mureka song be?

Mureka generates tracks up to 5 minutes long. For longer tracks, use the Extend feature to build on a generated song incrementally. Each extension continues from where the previous generation ended, matching the style and instrumentation.

Can I use Mureka-generated music commercially?

Yes, with the paid Basic plan ($8/month billed annually). The free trial does not include commercial licensing. The Basic plan includes commercial rights, making it safe for YouTube videos, podcasts, client projects, and other monetized content. Always check the current terms of service for your specific use case.

How does Mureka compare to Suno for lyrics-based songs?

Mureka's lyrics-first workflow is specifically designed for this use case. Its MusiCoT technology plans song structures around your lyrics before generating audio. Suno's Custom Mode also accepts lyrics, but the musical arrangement may not follow your structure tags as precisely. If writing lyrics first is your primary workflow, Mureka handles this better.

What audio formats does stem separation support?

Mureka accepts most common audio formats for stem separation. The output stems are provided as separate audio files that you can import into any DAW. The separation quality is solid and competitive with dedicated stem separation tools.

Can I use Mureka on my phone?

Yes. Mureka has native apps on both iOS and Android. You can generate songs, use melody idea input by humming directly into your phone's microphone, run stem separation, and access all major features. The mobile app is convenient for capturing melody ideas on the go and generating quick demos.