// News

What's actually happening in audio right now.

Our take on developments in plugin formats, sampling technology and audio tools — commentary, not a press-release rewrite. Every item links to its source; the analysis is ours. Cadmium Sampler's own release notes live in our newsletter, not here.

CLAP keeps gaining real hosts, not just enthusiasm

Bitwig and REAPER have supported CLAP at full depth for a while; FL Studio added it in 2024; Cubase has announced support with no firm date, and Ableton Live still hasn't committed either way. None of that reads as CLAP replacing VST3 — it isn't going to, not with fifteen years of VST3 momentum behind it — but the pattern is now a genuine three-format world rather than a curiosity format nobody ships.

The part worth understanding, not just the adoption count: CLAP's per-voice, non-destructive parameter modulation and its explicit thread-pool model solve real architectural limits in VST3, not cosmetic ones. Any sampler engine built around per-voice state — ours included — benefits directly from a host format that treats per-note modulation as a first-class citizen instead of a workaround. Source: ProducerGrid, "CLAP Plugin Format: Everything Music Producers Need to Know in 2026."

The Retromulator dispute is really a question every emulator has to answer

In March, discoDSP released a free plugin bundling several classic-synth emulations built on Gearmulator, an open-source project by The Usual Suspects that runs original hardware firmware to reproduce a chip's exact behaviour. discoDSP later donated $1,000 to The Usual Suspects; the money was sent back. Firmware from Access, Waldorf and Roland is copyrighted, was never licensed for this use, and the project depends on users supplying their own dumps rather than distributing ROMs directly — a gray area with no clean answer, not a case of one side being obviously right.

It's a useful contrast for anyone reverse-engineering old hardware: running someone else's firmware verbatim gets you exact behaviour with unresolved licensing; independently decoding a format from its bytes and verifying the result against real audio output is slower, but it's the only path that doesn't depend on a firmware dump you don't own the rights to. Source: Bedroom Producers Blog, "Why discoDSP Donated $1,000 to The Usual Suspects — And Got It Back."

The vintage-sampler sound is a full product category now, not a novelty plugin

UVI's Vintage Vault 5 gathers 43 instruments sampled from restored original hardware spanning five decades; their Emulation II+ set covers the Emulator, Emulator II and Emulator III lineage plus a dedicated drum machine built from Drumulator, SP-12 and E-mu samples. Separately, dedicated emulators of specific machines — the Yamaha TX16W, the Akai S900/S950 signal path — continue to find an audience precisely because they model one machine's quirks deeply rather than offering a generic "lo-fi" knob.

The throughline across all of it: the appeal isn't approximate nostalgia, it's the specific, measurable character of one real machine — its filter, its converter, its bit depth — captured and verified against the original rather than guessed at. That's the same bar we hold our own format decoders to. Sources: UVI, Vintage Vault 5, UVI, Emulation II+.

AI stem separation is now a workflow feature, not a specialist tool

Fender's Studio Pro DAW now integrates Moises stem separation and a "smart studio assistant" directly. Independent tests through early 2026 put newer architectures — Mel-Roformer, HTDemucs, MDX-Net — meaningfully ahead of the Spleeter-era tools that defined the category a few years ago, to the point that several reviewers now recommend a DAW's built-in separator before reaching for a third-party service at all.

Worth being precise about: stem separation is an offline, model-based reconstruction of audio that was never discretely available — a genuinely different problem from real-time sample playback, where the whole discipline is reproducing audio that already exists exactly, with nothing reconstructed or guessed. The two get lumped together as "AI in audio" more often than the underlying engineering has anything in common. Source: MusicRadar, stem separation tools tested, 2026.

Summaries above are written in our own words per item; anything in quotation marks is a short, attributed phrase from the linked source. Follow the links for the full articles. Want Cadmium Sampler's own release notes instead? Ask about our newsletter.