Monoidal categories are a powerful abstraction: examples of such algebras of processes include the computable pure functions of a language, queries to a database, or stochastic functions. Signal flow graphs can be interpreted in each one of these categories.
Stream-based programming, on the other hand, is in principle restricted to the case of plain functions. Devising stochastic, relational or quantum versions of a stream-based programming language needs additional work if we do not posess a good notion of stream in a monoidal category.
The present work tries to fill this gap: we provide a notion of synchonous stream transformer in an arbitrary monoidal category and use it to give semantics ot a stochastic dataflow language.