Convertra — DJ audio platform

A native macOS app for DJs — and the on-device engine that powers it.

Convertra is two products, one brand. Convertra App organises, analyses and converts a DJ’s whole library on macOS. Convertra AudioCore is the on-device key, tempo and Camelot engine inside it — validated against Mixed In Key on 1000+ commercial tracks, and available to license.

Native macOS app · Apple Silicon + Intel · AudioCore engine · ~336 KB, zero runtime dependencies

~70%
Exact Camelot key match
~83%
Harmonic / mix-compatible match
~82%
Tempo within ±0.5 BPM
~1 s
Analysis per track (release build)
100%
On-device — no cloud calls

Benchmarked against the same class of technology that powers

Mixed In Key Rekordbox

Comparison names are referenced for category context only. Convertra is independent and not affiliated with or endorsed by these products.

One platform, two products

Buy the app. License the engine.

Convertra ships as a finished macOS app for DJs and as an embeddable engine for developers — same brand, same technology, two ways to work with it.

For DJs

Convertra App

A native macOS library, analysis and conversion app — key, BPM, Camelot, metadata, duplicates and lossless-to-MP3 in one workflow.

Get Convertra App
For developers & companies

Convertra AudioCore

The on-device DSP engine behind the app — musical key, tempo and Camelot detection, delivered as a signed universal binary to embed, white-label or license.

License AudioCore
Technology

A full analysis pipeline, not a wrapper

Every stage — decode, onset, pitch-class profiling, autocorrelation, Camelot mapping — is native Swift DSP written from the ground up. No FFmpeg in the analysis path, no external model weights.

Musical Key Detection

Peak-picking harmonic pitch-class profiling (HPCP) over a resampled mono signal, resolving tonic and mode. Outputs the musical key plus mix-compatible candidates with confidence.

BPM Detection

Multi-band onset-envelope extraction feeding integer-lag autocorrelation, refined by parabolic peak interpolation for continuous, sub-BPM precision — not quantised guesses.

Camelot Conversion

Detected keys map directly to Camelot / Open Key notation, the harmonic-mixing standard DJs actually use — colour-coded and ready to drive compatibility logic in your app.

Audio Processing Pipeline

Single-pass fast decode of the track core via AVAssetReader to 22.05 kHz mono, then key and tempo detection run concurrently per track. Deterministic, reproducible output.

Performance Architecture

Built on Apple Accelerate/vDSP with actor-based concurrency. Release-optimised binary analyses a track in roughly a second — about 10× faster than the debug-compiled source.

Delivery Model

Shipped as a signed, closed-source .xcframework — a ~336 KB universal binary you drop into a Swift/Objective-C project. The public adapters preserve stable method signatures.

Performance & honesty

Numbers we can reproduce

These are measured results from a from-scratch DSP rewrite, validated against Mixed In Key ground-truth tags over 1000+ real commercial tracks (house, tech-house, hip-hop, pop) — not marketing figures.

Harmonic / mix-compatible match ~83%
Tempo within ±0.5 BPM ~82%
Exact Camelot key match ~70%

How it was measured

A tuning harness compares AudioCore output track-by-track against a local Mixed In Key CSV export. Key is scored on exact Camelot match and on harmonic (mix-compatible) match; tempo on the share of tracks within ±0.5 BPM. Speed is end-to-end per track on a release build.

1000+ tracks Mixed In Key ground truth ~1s / track · release

Fully on-device

No network round-trips, no per-track API cost, no data leaving the machine — analysis scales with local CPU, not a cloud bill or rate limit.

Built for large libraries

Concurrent per-track analysis and bounded processing keep it stable across tens of thousands of files — the same engine that drives the Convertra desktop app.

Native, not wrapped

No Electron, no web runtime, no cloud analysis dependency. Apple-native audio all the way down, shipped as a single universal binary.

Convertra App

The macOS app DJs run every day

A finished product in its own right: Convertra organises, analyses and converts a DJ’s entire library on macOS — key, tempo and Camelot on every track, metadata and artwork editing, duplicate detection and lossless-to-MP3 conversion. Powered by AudioCore, built for working DJs.

Get Convertra App
Convertra — library view with on-device key/BPM/Camelot analysis
Convertra — library view with on-device key/BPM/Camelot analysis
Convertra — metadata & artwork editor
Convertra — metadata & artwork editor
Convertra — batch conversion queue
Convertra — batch conversion queue
Convertra — duplicate detection across the library
Convertra — duplicate detection across the library
Convertra — conversion & library settings
Convertra — conversion & library settings
01

Library & analysis

Ingest thousands of tracks, each analysed for key, BPM and Camelot on-device. Colour-coded harmonic keys, sortable and searchable.

02

Metadata & artwork

mp3tag-grade tag editing with a native ID3v2 writer that embeds cover art into MP3 and AIFF — including formats FFmpeg silently drops.

03

Conversion pipeline

Lossless-to-MP3 batch conversion with a bounded-concurrency queue, sample-rate and channel options, and folder-structure preservation.

Category context

Where AudioCore sits

A structural comparison of how key/tempo analysis is delivered across common tools. Feature availability in third-party products is summarised for orientation and may change.

Capability Convertra AudioCore Mixed In Key Rekordbox
100% on-device DSP
Licensable as an embeddable SDK
White-label / OEM integration
Native Camelot / Open Key output
No cloud, no neural-net weights ~ ~
Universal binary footprint ~336 KB n/a Bundled

Convertra AudioCore is designed to be embedded in your product. Mixed In Key and Rekordbox are end-user applications — the comparison highlights delivery model, not a claim of superiority on every axis.

Opportunities

Five ways to work with Convertra

The technology, the product and the brand are structured so a partner can license a component, integrate under their own brand, co-develop, invest, or acquire the whole thing.

Acquisition

Full acquisition of the project — source code, AudioCore IP, brand, documentation, build/release infrastructure and intellectual property in one transfer.

Discuss acquisition

Licensing

Embed AudioCore as a drop-in key/BPM/Camelot SDK. SDK, Commercial and Enterprise tiers to match your distribution and support needs.

License AudioCore

White Label

Ship AudioCore analysis under your own brand inside a DJ app, library manager or music platform — same engine, your product surface.

Explore white-label

Partnerships

Joint development, strategic and technology partnerships — from a co-built analysis feature to a deeper platform collaboration.

Start a partnership

Investment

Investment, equity participation or development funding to accelerate the roadmap and expand AudioCore beyond macOS.

Investment inquiry
Founder
Hamid Kazimov

Hamid Kazimov

Founder · Software Developer · Creator of Convertra & Convertra AudioCore · DJ & Music Director

Convertra AudioCore was built by a working DJ and software developer — someone who needed commercial-grade key and tempo analysis that runs locally, and rebuilt the DSP from scratch when off-the-shelf options fell short. The result is an engine validated against the tools professionals already trust, honest about what it does, and engineered as native Apple audio from decode to Camelot.

FAQ

Questions, answered

What exactly is Convertra AudioCore?

A proprietary DSP engine that detects musical key, tempo (BPM) and Camelot notation from audio, entirely on-device. It powers the Convertra macOS app and is available to license as an embeddable component.

How accurate is it?

Measured against Mixed In Key ground-truth tags on 1000+ commercial tracks: ~70% exact Camelot key match, ~83% harmonic (mix-compatible) match, and ~82% of tracks within ±0.5 BPM. These are reproducible numbers, not marketing claims.

Does it use the cloud or a neural network?

No. There are no cloud calls, no third-party analysis API and no neural-network weights. It is pure DSP built on Apple’s Accelerate/vDSP, running 100% locally.

How is it delivered and integrated?

As a signed, closed-source .xcframework — a ~336 KB universal binary (Apple Silicon + Intel) that drops into a Swift or Objective-C project. Public adapters expose stable method signatures for key and tempo results.

What platforms are supported today?

The shipping engine targets macOS (12+) as a native binary. Porting and expansion beyond macOS is exactly the kind of roadmap a licensing, partnership or investment conversation can prioritise.

What licensing options are available?

SDK, Commercial and Enterprise licensing, plus white-label/OEM integration and full acquisition. The right structure depends on your distribution model — reach out and we’ll scope it.