> For the complete documentation index, see [llms.txt](https://musenai.gitbook.io/musenai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://musenai.gitbook.io/musenai/understanding-musen/ai-radio-vision-paper.md). # AI Radio Vision paper A Vision for Listening\ After Access ------------ ### Vision Paper | August 2026

Access gave us abundance. Abundance made
orchestration scarce.

AI Radio is the idea that intelligent systems can shape continuity around human culture without requiring the listener to keep choosing, and without requiring AI to replace the culture being heard.

*This paper presents the cultural thesis behind AI Radio. Formal definitions, system properties, evaluation methods, limitations and research questions are addressed in the musen White Paper.*

ACCESS • ABUNDANCE • ORCHESTRATION


THE HISTORICAL TRANSITION 1\. After access Streaming achieved something extraordinary. It transformed music from a scarce physical or local resource into a global service. For hundreds of millions of people, an enormous catalogue can now be reached in seconds. In 2025, global recorded music revenue reached US$31.7 billion, and paid subscription accounts represented 837 million users worldwide. Access at unprecedented scale is largely established, even though rights, geography, affordability and connectivity still create real limits. \[1] That success changes the question. When access is scarce, the problem is finding the work. When access is abundant, the problem increasingly becomes deciding what deserves attention now, what should follow, and how an experience can remain coherent without demanding constant management from the listener.
Access gave us everything. It did not tell us what should happen next.
This is not an argument against search, playlists, albums, recommendation, or explicit control. Those forms remain valuable. The claim is narrower: explicit choice should not be required at every moment of listening. A person working, driving, cooking, moving through a city, resting, or talking with friends may want music deeply while wanting very little interface. Abundance therefore creates a new scarcity. The scarce capability is no longer merely distribution. It is orchestration: turning a huge possibility space into a coherent moment without flattening culture into a single optimization target. | VISION THESIS | As availability expands, the value of timing, sequence, context and editorial judgement becomes more visible. | | ------------- | ------------------------------------------------------------------------------------------------------------- |
THE HUMAN FORM OF THE MEDIUM 2\. Listening is lived in time Music is not only an object people select. It is something people live alongside. Research on everyday music listening shows that listening behaviour changes with device, time of day, context, mood, attention and perceived control. Radio research also describes listening that can be simultaneously background, routine and companionship. These are not edge cases. They reveal that the value of audio often lies in how it accompanies life rather than in how often a person actively chooses the next item. \[2] \[3] Radio has always understood this intuitively. A radio experience is made from individual works, but it is not reducible to them. Sequence matters. Pacing matters. Recurrence matters. Contrast matters. Speech, silence, familiarity and surprise can all change the meaning of what follows. The experience unfolds through time.
A catalogue is a library. Radio is time.
That distinction is cultural before it is technical. A library maximizes access to discrete works. Radio takes responsibility for continuity. It creates a present tense: something is happening now, and what happens next is shaped by what came before. Traditional radio achieved this through human programming, format clocks, presenters, schedulers and editorial judgement. AI Radio asks whether computational systems can extend that temporal intelligence without removing human culture from the centre. The ambition is not to make every listening moment passive. It is to make active choice optional when continuity is more valuable than control. |

NOT A FALSE
CHOICE

| Choice and continuity are not opposites. AI Radio argues that listeners should not need to exercise explicit control continuously in order to receive a coherent experience. | | ---------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
A DIFFERENT QUESTION 3\. From recommendation to orchestration Recommendation asks which item is likely to be relevant. Orchestration asks how an unfolding experience should continue. The difference is subtle but important. A ranked track can be individually appropriate and still be wrong for the moment, the sequence, the energy curve, the amount of speech already heard, or the emotional direction of the session. The emerging industry already points toward intent as a more natural interface. Spotify now lets eligible listeners shape personalised sessions through DJ requests and conversational controls. YouTube Music lets eligible users describe a mood, activity or listening experience and generate a customised station or mix. These developments do not prove demand for musen, and they do not make the systems equivalent. They show that the interface to music is expanding beyond browsing and manual construction. \[4] \[5] \[6]
Recommendation chooses an item. Radio shapes a sequence.
AI Radio takes this movement seriously enough to make continuity the primary product objective. The listener can express intent, but the system carries responsibility for what happens between interventions. It can observe the evolving session, preserve relevant state, and adapt future editorial decisions as the context changes. The important shift is therefore not from human control to machine control. It is from continuous management to continuous support. The listener should be able to intervene when they care, then return attention to life while the experience continues coherently. |

THE CORE
QUESTION

| What should be happening now, and how should the experience continue? | | --------------------------- | --------------------------------------------------------------------- |
A CULTURAL CHOICE ABOUT AI 4\. AI in service of culture Generative AI may become a legitimate creative medium in its own right. musen is making a different choice. It applies AI primarily to orchestration rather than music generation. The current musen product does not include AI generated music. That choice matters because the supply problem is changing quickly. Deezer reported in July 2026 that it was receiving about 90,000 fully AI generated tracks per day, with such tracks exceeding half of daily new deliveries at peak. Yet the same company reported that fully AI generated music represented only a small share of listening on its service. Supply can expand much faster than attention. \[7]
AI does not need to replace culture to transform listening.
For musen, the more interesting role for AI is to help human work remain encounterable in conditions of abundance. It can interpret intent, preserve continuity, manage pacing, notice when the session should change direction, and help the listener move through existing culture without turning every moment into a search task. This requires memory, but memory should not become an unrestricted accumulation of a person. A radio may remember enough to evolve, while keeping that state governed, correctable, limited in purpose and subject to forgetting. Personalisation should serve continuity, not create captivity.
Memory should create continuity, not captivity.
The same principle applies to agency. An intelligent radio should become easier to guide as it learns, not harder to escape. Retune, explicit requests, correction and the ability to change direction are not failures of automation. They are part of respectful automation.
RECOGNITION IN CONTINUOUS MEDIA 5\. Attention, attribution and value If an experience unfolds through time, then time becomes an important fact about what actually happened. A listener did not merely trigger an item. They spent minutes inside a sequence, heard particular works, encountered a particular editorial structure, and remained or left. This does not mean duration is the same as satisfaction, fairness or economic value. Listening time can include distraction, habit, interruption or passive exposure. But it is a richer primitive than a single click when the medium itself is continuous. That opens a design question for creator economics. If radio is experienced through time, creator recognition may need to become more sensitive to time, context and editorial contribution. musen is exploring this through Checkmate, a planned research and design direction for attributable listening and future economic allocation. Current creator payouts are not live, and no claim is made that duration alone provides a complete or fair economic model. | DESIGN PRINCIPLE | Time can inform attribution. It should not be mistaken for a complete measure of human value. | | ---------------- | --------------------------------------------------------------------------------------------- | Attribution also depends on provenance. As AI becomes more involved in cultural systems, works, contributors and permissions should become more legible rather than less. Signed provenance frameworks can bind verifiable assertions to assets, while independent timestamping can provide evidence that a particular data commitment existed by a certain time. These mechanisms can strengthen evidence, but they do not prove authorship, ownership, truth or consent by themselves. \[8] \[9] The deeper principle is simple: intelligent systems should not make culture more anonymous. A future radio should be able to know what it is allowed to use, who contributed to what is heard, what was actually experienced, and which claims are evidence rather than assumption.
ONE ATTEMPT TO MAKE THE IDEA REAL 6\. musen’s attempt musen is building AI Radio as a real listening system rather than treating the idea as a speculative interface concept. As of August 2026, the public product is a free Live Radio experience that requires an account. An AI Host may speak within Live Radio. The AI DJ is a silent orchestration engine, but it is not the current public mobile product. The product does not currently include AI generated music. musen is developing the system around a simple product feeling: listening should be effortless, personal, alive and adaptive. The listener can describe a moment and then stop managing the experience. The radio should continue.
The world can provide context. musen’s job is to shape the radio.
Several ideas discussed around musen remain future directions rather than current capabilities. Time aware creator economics, broader provenance infrastructure, public developer interfaces, machine controlled contexts and richer forms of distributed intelligence all require further technical, legal and economic validation. They belong to the horizon of the category, not to a claim that the product is already complete. Over time, the interface to radio may become less important than the continuity of the radio itself. A car, home, game or software agent may understand something about the surrounding moment. That environment can express context or intent, while the radio preserves its own editorial continuity. The screen can become one interface among many, but the human experience remains the reason for the system. DECLARATION Access gave listeners an extraordinary catalogue of human expression. It did not remove the need for attention, judgement, timing or care. As music becomes more abundant and systems become more intelligent, the central challenge is not merely to retrieve another relevant item. It is to shape continuity without taking agency away, to use memory without creating captivity, and to apply AI without making creators disappear. That is the possibility of AI Radio: not less human culture, but a more living way to encounter it.
The future of audio may not be more things to choose. It may be an experience that knows how to continue.

EVIDENCE AND CONTEXT 7\. References and evidence notes This Vision Paper is normative and philosophical. The sources below establish context for the historical transition, everyday listening behaviour, current product directions in music services, rapid growth in generated supply, and provenance technology. They do not constitute proof that AI Radio will succeed or that musen has established product market fit. \[1] IFPI. Global Music Report 2026, State of the Industry. March 2026. [Source](https://www.ifpi.org/global-music-report-2026-global-recorded-music-revenues-grow-6-4-as-record-companies-drive-innovation/) \[2] Krause, Amanda E., Adrian C. North, and Lauren Y. Hewitt. Music listening in everyday life: Devices and choice. Psychology of Music, 2015. [Source](https://journals.sagepub.com/doi/10.1177/0305735613496860) \[3] Krause and colleagues. Research on the role and impact of radio listening practices in older adults’ everyday lives. Frontiers in Psychology, 2020. [Source](https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2020.603446/full) \[4] Spotify. Your Personalized DJ Experience Is Expanding With 4 New Languages and More Markets. May 7, 2026. [Source](https://newsroom.spotify.com/2026-05-07/dj-expansion-4-new-languages/) \[5] Spotify. Just Say the Word: A More Personal Way to Ask, Discover, and Listen. July 14, 2026. [Source](https://newsroom.spotify.com/2026-07-14/talk-to-spotify-announcement-beta/) \[6] YouTube Music Help. Ask Music for music and podcasts. Accessed August 2026. [Source](https://support.google.com/youtubemusic/answer/17090260?hl=en) \[7] Deezer. AI music has surpassed 50 percent of new music uploads for the first time. July 21, 2026. [Source](https://newsroom-deezer.com/2026/07/ai-music-exceeds-50-percent-daily-uploads-deezer/) \[8] Coalition for Content Provenance and Authenticity. Content Credentials specifications and guiding principles, current specification family 2026. [Source](https://spec.c2pa.org/specifications/) \[9] OpenTimestamps. Timestamping proof standard and documentation. Accessed August 2026. [Source](https://opentimestamps.org/) |

COMPANION
PAPER

| The musen White Paper addresses formal definitions, system properties, evaluation, limitations, falsification and open research questions for AI Radio. | | ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- | *AI Radio is presented here as a cultural and design thesis. Claims about current musen product status are dated August 2026 and should be updated when the product changes.*