The industrialization of perception: MusiMap, Tuned Global and the 200-million-track bottleneck
In September 2026, two apparently distant announcements told the same story from two different points in the music infrastructure. MusiMap returned to the market with a suite capable of analyzing, searching and profiling music through intelligence built on nearly forty years of musicological research, while Tuned Global increased its ingestion capacity from 2 million to 20 million tracks per day in order to absorb catalogs that can now arrive in batches of 30 to 60 million tracks. As scale exceeds what human intervention can directly navigate, the bottleneck begins to move. Transporting 200 million tracks is no longer enough, they must also be made intelligible quickly enough to remain genuinely exploitable.
Klem Loden
9/16/20264 min read
When one million tracks teach the machine how to listen
MusiMap has come a long way. Founded in 2008 by Belgian musicologists from research dating back to 1986, acquired by Utopia Music in 2021 and later returned to independence following the difficulties of its former parent company, the business unveiled a new architecture on September 10 built around four products, MusiTag for track analysis and enrichment, MusiSearch for semantic search, MusiProfile for audience profiling and MusiTrace for audio identification and monitoring.
The technology obviously matters, yet MusiMap’s claimed distinction lies almost upstream of it. More than one million tracks have been manually analyzed by musicologists, creating a foundation from which automation can learn to recognize genre, mood, context and sonic characteristics with greater granularity than conventional descriptive metadata. The arrival of Ty Roberts, co-founder of Gracenote and former CTO of Universal Music Group, as Chief Commercial Officer now accompanies the commercialization of that intelligence at a larger scale.
Light On became its first announced commercial client. The global music strategy and brand partnerships agency plans to integrate the suite into its workflows to identify music corresponding to a brand’s identity, audience and desired emotional impact. The case is particularly revealing because music search gradually moves beyond asking what a track is and begins asking what it might mean within a given context, for a given audience.
In other words, the machine no longer needs merely to retrieve the music... it needs to begin understanding it well enough for someone to retrieve it for the right reason.
When the catalog arrives by the tens of millions
At the other end of the problem, Tuned Global is not attempting to understand an individual track more finely, it is trying to prevent sheer volume from becoming an obstacle in itself.
On September 11, the company detailed the modernization of its Music Cloud Platform infrastructure with AWS and SourceFuse. The migration from Microsoft SQL Server to Amazon Aurora PostgreSQL, accompanied by a new ETL architecture relying notably on AWS Glue and Amazon EventBridge, increased onboarding ingestion capacity from 2 million to 20 million tracks per day. Large catalogs that could previously require up to two weeks can now be integrated in one to two days, while query performance has improved three to five times and infrastructure costs have fallen by 33%.
Behind those performance figures lies a far more interesting measure of the change in scale. Tuned Global currently manages approximately 200 million tracks and says new clients can arrive with 30 to 60 million tracks that need to be ingested, standardized, enriched with metadata and then made available through more than 500 APIs. At this level, the catalog ceases to be merely a collection of assets to administer, it becomes a flow the infrastructure must be capable of absorbing without allowing its volume to slow the exploitation of what it contains.
The approximately 25% reduction in human intervention observed by Tuned Global within its own ingestion and scaling workflows then takes on its proper meaning. It does not describe the disappearance of human expertise, it measures what happens when certain repetitive operations cease to require it so that scale can continue to grow.
The real bottleneck is moving
Taken separately, MusiMap and Tuned Global solve two different technical problems. Their temporal proximity nevertheless allows the same tension to be observed from opposite ends of the pipeline.
Tuned Global is working on a system’s ability to absorb, normalize and serve a musical mass that has become enormous. MusiMap is working on another system’s ability to produce enough intelligence about the content for an individual track to remain distinguishable within that mass. One accelerates the movement of the catalog, the other increases the amount of exploitable meaning attached to what passes through it.
This is precisely where volume changes the nature of the problem. When a few thousand tracks need to be explored, human expertise can still directly navigate a significant portion of the catalog. When the infrastructure contains approximately 200 million and 20 million can be ingested in a single day, that relationship becomes mathematically impossible to maintain in the same form. The question is therefore no longer simply how much music a platform can store, but how much of it it can make effectively searchable.
MusiMap presents a particularly interesting paradox in this respect. The automation it commercializes rests precisely on a massive accumulation of human judgment, nearly forty years of research and more than one million tracks manually analyzed. The industrialization of musical perception therefore does not necessarily imply the disappearance of expertise, it may also indicate its displacement upstream, where it becomes the reference material that subsequently allows the machine to operate at a scale no team of musicologists could navigate track by track.
From the existence of a catalog to its exploitability
For synchronization, this transformation deserves close attention because discovery occurs well before clearance. A track may have perfectly administered rights and a technically available master, yet it still has to emerge at the precise moment when a search calls for an emotion, a context, an energy, an era, a texture or a combination of criteria that conventional metadata describes imperfectly.
As catalogs grow, the quality of this layer of understanding could therefore weigh more heavily on their exploitability. It would be premature to conclude that an insufficiently enriched asset automatically becomes invisible to the market, since neither MusiMap nor Tuned Global establishes such a causal relationship. Their simultaneous evolution nevertheless shows an industry investing in its ability to interpret and interrogate the mass alongside its ability to transport it.
The bottleneck has therefore not disappeared with increased technical capacity, it has simply moved. An infrastructure capable of ingesting 20 million tracks per day solves one part of the volume problem, then immediately exposes the next question... among everything that has just entered, how do you find what you are looking for?
Perhaps this is where the industrialization of perception truly begins.
Verified Sources and References
Music Ally, “MusiMap unveils its revamped suite of ‘music intelligence’ tools,” September 10, 2026.
Record of the Day, “MusiMap launches latest technology unlocking new commercial opportunities, alongside first deal and appointment of Ty Roberts as commercial lead,” September 11, 2026.
Tuned Global, “Tuned Global scales AI-ready music platform foundation with AWS and SourceFuse,” September 11, 2026.
TechPartner News, “Tuned Global scales music platform with AWS and SourceFuse,” September 15, 2026.