EmergingSpeech, audio and music intelligence

Music Generation Researcher

Also called Music AI Researcher

A Music Generation Researcher builds music models that can be steered by musical constraints and checks whether their output copies existing work. It matters because a model people cannot control or clear for rights is hard to ship.

What they do

The work starts with how music is represented: audio codecs, symbolic formats, stems, tempo and key. From there they train and adapt sequence models, often Transformers, to generate music that follows instructions such as structure, instrumentation, mood or a given melody. They build evaluation as they go. That means setting up listening tests, recruiting suitable listeners, and pairing those results with automatic checks for key, tempo and form. They also test outputs against known works to look for memorisation and close resemblance, and write up what they find for product and legal colleagues.

What a strong hire delivers

In the first months, a strong hire delivers a working baseline model and a clear account of how well it follows musical constraints. They set up a repeatable listening-test process that the team trusts, and the first rights and originality assessment, with plain findings on where the model copies and where it does not. They also leave the team with a shortlist of next experiments, ranked by what listeners actually noticed.

How you can tell they are good

Ask them to walk through one generation system they built, covering the representation choice and why they made it. Ask how they would tell whether a model is following a constraint or merely sounding plausible. Good candidates describe listening tests with care: who listened, what was asked, how bias was handled. They talk about originality risk with specifics. Be wary of anyone who relies only on automatic metrics, or who treats rights as someone else's problem.

Why it is hard to fill

Hardly anyone holds this title. The work often sits inside speech, audio or general ML research jobs, so searching by title finds few people. The mix is rare: serious ML skill, real musical understanding and comfort with rights questions. Strong candidates are often happy where they are, and many lean toward one side, either musicians who code or engineers who rarely listen. Expect to assess for the missing half.

Where to find them

Look in speech processing and acoustics groups, music technology research labs and university audio departments. Some come from ML teams at frontier labs or model providers working on audio generation. Others come from music software, streaming and creative tools companies, where they worked on music retrieval or synthesis. Titles vary: audio researcher, research scientist, applied scientist, DSP engineer, music information retrieval researcher. Active musicians with ML training are worth a conversation.

Indicative pay

Not enough published evidence yet to give a fair range for this role, which is common for work this new. We benchmark it for you as part of a search.

Interview questions for a Music Generation Researcher

  1. Tell us about a music or audio generation model you built. What representation did you choose and why?

    A strong answer names the representation, the trade-offs against alternatives, and what broke along the way. Listen for ownership of decisions. Worry if they describe only using someone else's pipeline or cannot explain the choice.

  2. How would you measure whether a model follows a requested key, tempo or structure?

    They should combine automatic checks, such as detecting key and tempo, with human judgement for form and feel. They should know where automatic checks mislead. Worry if they offer a single metric as the answer.

4 more questions, with what a strong answer covers, are in the Role Kit.

Job description you can post

We are building music generation models that people can direct and that we can stand behind. We are looking for a researcher who understands both the machine learning and the music. You will work with engineers, product and legal colleagues to get from experiment to something we can release.

What you will do

  • Develop generative models for music with controls such as structure, instrumentation and style.
  • Choose and test music representations for training and generation.
  • Design and run listening tests and pair them with automatic musical checks.
  • Assess outputs for memorisation and resemblance to existing works.
  • Write clear findings on quality, controllability and rights risk.
  • Work with engineers to move successful experiments into production.
  • Keep the team informed on relevant research in audio and music modelling.

What you will bring

  • Demonstrated work building generative models for audio or music.
  • Strong grounding in deep learning and sequence models such as Transformers.
  • Experience with audio toolchains and music or speech representations.
  • Experience designing human evaluations or listening tests.
  • Musical understanding that shows in your technical choices.
  • Clear written communication with non-researchers.
  • Sound judgement about originality and rights questions.

Nice to have

  • Background in acoustics, speech processing or music technology.
  • Experience with symbolic music data alongside audio.
  • Published work, demos or open projects in music generation.

Free to use and adapt. If it saves you time, a link back to this page is appreciated.

Skills that matter

  • Speech and audio modellingAdvancedRecognise, synthesise or analyse audio
  • Deep learningAdvancedBuild and diagnose trainable neural models
  • Linguistics and localisationWorkingAdapt language behaviour across cultural and linguistic contexts
  • Model evaluationWorkingMeasure capability and failure modes credibly

Other roles in speech, audio and music intelligence

Hiring nowSpeech Recognition EngineerA Speech Recognition Engineer builds the system that turns spoken language into accurate text for real users in real conditions. It matters because…Hiring nowSpeech Synthesis ResearcherA speech synthesis researcher builds text-to-speech models that sound natural, stay intelligible and respect speaker consent. Companies need them to…Hiring nowAudio Intelligence ScientistAn Audio Intelligence Scientist builds models that recognise sound events and reason about acoustic scenes. They make audio features dependable in…

Related roles in other families

Hiring nowPost-Training Research EngineerPost-training and preference optimisationA Post-Training Research Engineer adapts pretrained models into ones that behave as the company needs. They turn preference data and reward signals…Hiring nowReward Modelling ResearcherPost-training and preference optimisationA reward modelling researcher builds models that learn what good output looks like from preference comparisons. They give post-training teams a…Hiring nowReinforcement Learning Research EngineerPost-training and preference optimisationA Reinforcement Learning Research Engineer improves how a model behaves by training it against interactive signals. They matter because reward gains…Hiring nowNatural Language Processing ScientistLanguage and multilingual intelligenceAn NLP Scientist builds and tests methods for language understanding and generation. They make sure a company's language models work across…

How we wrote this

Role definitions are our own editorial work, checked against occupational classifications, standards, research and employer pages. Titles vary by company, and this is a description of the work, not a job advert.

← Speech, audio and music intelligenceAll new job families

Start a search

Hiring a Music Generation Researcher?

Tell us the work. We find the people already doing it, whatever their job title says, and you meet them already interviewed for your role.