Flint Image 4K is here. Frontier quality, at a fraction of the price.
Mureka Stem Song API image preview

Mureka Stem Song API

Text·mureka·by Mureka

Mureka Stem Song is a music processing model that separates a song into individual audio components such as vocals and instruments.

Try it now →
API reference
Runtime (p50)
10s
Estimated price
$0.1
Call the API
prediction.sh
sh
curl -X POST \
  -H "Authorization: Bearer $EACHLABS_API_KEY" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "mureka-stem-song",
    "version": "0.0.1",
    "input": {
        "url": "https://storage.googleapis.com/magicpoint/outputs/mureka_stem_song_input.mp3"
    },
    "webhook_url": ""
}' \
  https://api.eachlabs.ai/v1/prediction/
Documentation8 sections
  • Overview

    mureka-stem-song — Audio Stem Separation AI Model

    Mureka Stem Song, part of the Mureka family from provider Mureka, revolutionizes music production by separating full songs into individual audio stems like vocals, drums, bass, and other instruments with high precision. This mureka-stem-song model enables producers and remixers to isolate components for remixing, editing, or professional mixing without complex manual processing. Developed using advanced AI like MusiCoT technology, it stands out for delivering clean stem exports compatible with digital audio workstations (DAWs), making it ideal for users searching for "AI stem separation tool" or "song stem splitter AI".

  • Capabilities
    • Separates songs into stems like vocals, instruments, drums, and bass with professional fidelity
    • Generates complex musical structures using MusiCoT for coherent arrangements across genres
    • Produces realistic vocals and instrumentation, rivaling high-end production tools
    • Supports instrumental-only tracks with customizable mixing controls
    • Handles voice cloning and multi-speaker speech integration seamlessly
    • Exports high-quality audio in multiple formats, including isolated stems for remixing
  • Use cases

    Use Cases for mureka-stem-song

    Music producers use mureka-stem-song to deconstruct reference tracks for sampling, uploading a full song and extracting clean drum and bass stems for new beats in their DAW. This saves hours of manual editing, enabling quick iteration on remixes for electronic or hip-hop projects.

    Developers integrating "AI audio stem separation API" into apps feed user-uploaded songs into the mureka-stem-song model, automatically generating isolated vocals for karaoke tools or lyric video creators. For example, input a pop track with the parameter "separate vocals and instruments," yielding editable stems for real-time processing.

    Content creators and DJs apply it for live performance prep, separating instrumentals from vocals in live recordings to build custom mashups. Marketers producing ad music isolate elements like guitars for branded jingles, streamlining workflows for "song to stems AI" needs.

    Sound designers for film and games extract ambient instruments from orchestral pieces, using the stems to layer into immersive soundscapes without original multitracks.

  • Tips & tricks

    How to Use mureka-stem-song on Eachlabs

    Access mureka-stem-song seamlessly through Eachlabs Playground for instant testing—upload your audio file, select stem types like vocals or drums, and generate separated WAV tracks in seconds. Integrate via the Mureka API or SDK for apps, specifying input audio and output formats for high-fidelity stems at 44.1 kHz, perfect for production pipelines.

    ---

  • Technical spec

    What Sets mureka-stem-song Apart

    The mureka-stem-song model excels in precise stem separation, breaking down tracks into vocals, instruments, and more while preserving audio quality at studio-grade levels. This allows users to export up to multiple time-aligned WAV stems for seamless import into DAWs like Ableton or Logic, a feature tailored for professional workflows.

    • Individual Stems Export: Generates separated tracks for instruments and vocals directly from input audio. This empowers advanced mixing and remixing without bleed between elements, perfect for "AI music stem separator" applications.
    • High-Fidelity Processing: Supports 44.1 kHz quality with clear instrument separation, even in complex arrangements. Users gain professional-grade outputs ready for commercial use, differentiating it from basic splitters.
    • Integration with Mureka Ecosystem: Leverages voice cloning and reference audio upload for guided separations. This enables customized stem isolation based on specific vocal timbres or instrumental references, ideal for "Mureka stem song API" developers.

    Technical specs include input of full audio files (MP3/WAV), output in WAV stems, and rapid processing in seconds, supporting tracks up to several minutes long.

  • Things to be aware of
    • Stems download and instrumental generation added in March 2025 with O1 model, improving over V5.5/V6
    • V7.5 default since September 2025 offers better quality and longer prompts than prior versions
    • Users report strong genre versatility and audio fidelity, especially for realistic vocals
    • Resource needs scale with concurrency; up to 10 tracks supported, unlimited on Ultra plans
    • Positive feedback on MusiCoT for advanced compositions and stem tools as a producer's toolkit
    • Common concerns include fallback to default engines in TTS and deprecated older endpoints
  • Key considerations
    • Use detailed prompts with genre, mood, structure, and lyrics for best musical coherence
    • Prioritize recent models like V7.5 or O1 for superior stem quality and vocal realism
    • Account for increased character limits (1000 for prompts, 3000 for lyrics) to enable complex inputs
    • Balance generation speed with quality by selecting appropriate models; O1 offers advanced structures but may require more processing
    • Avoid exceeding rate limits on non-unlimited endpoints; speech generation has no limits
    • Test custom vocals with clean MP3 samples to prevent cloning artifacts
  • Limitations
    • Older models like V6 deprecated in favor of V7.5/O1, potentially limiting legacy compatibility
    • Dependent on clean input audio for accurate stem separation; noisy uploads may degrade results
    • No public benchmarks on exact separation accuracy or real-time processing speeds

Related models

4 models
Stable Audio 2.5 · Text to Audio AI model preview
Stable Audio 2.5 · Text to AudioStability
Mureka Generate Track · Stem Generation AI model preview
Mureka Generate Track · Stem GenerationMureka
Mureka · Generate Remix AI model preview
Mureka · Generate RemixMureka
Kokoro 82M AI model preview
Kokoro 82MKokoro
* FAQ

About Mureka Stem Song API

01 / 03

What is Mureka Stem Song?

Mureka Stem Song is an AI music model by Mureka that generates songs with separated audio stems individual tracks for vocals, drums, bass, and other instruments. This enables flexible post-production mixing and remixing of AI-generated music within standard audio production workflows.