CH-03 · Architecture
Audio environment comparison
This is an independent, technical comparison of four widely used audio environments — Audacity, Ocenaudio, Ardour and Reaper. The desk judges them only by engineering characteristics: how each one models an edit, what number format it works in, how it meters a signal, and how deep its routing goes. No brand is endorsed, no interface is reproduced, and no product files are offered here.
Names are used only to identify the objects compared, under nominative fair use, to identify the objects being compared. Nothing on this page is affiliated with, sponsored by or approved by any of the four projects or their publishers, and no logo, screenshot or copied interface element appears in this reference.
Method
Each environment was assessed against the same five questions, answered from its published engineering behaviour rather than from marketing language. The aim is a compact matrix a reader can scan before choosing a working desk, not a ranking. A tool that suits a restoration desk may be a poor fit for a session that lives in a deep routing graph, and the reverse is equally true.
| Characteristic | Audacity | Ocenaudio | Ardour | Reaper |
|---|---|---|---|---|
| Edit model | destructive regions, saved as copies | region edits with cached undo | non-destructive graph | non-destructive graph |
| Working format | 32-bit float internally | 32-bit float internally | 32-bit float internally | 64-bit float internally |
| Metering | sample peak, basic readout | sample peak plus spectrum view | peak, RMS and loudness | peak, RMS, loudness, true peak |
| Routing depth | single bus, limited sends | single bus, no buses | buses, inserts, side-chain | buses, inserts, multichannel |
| Restoration set | spectral repair and de-hum | spectral repair and effects racks | relies on external nodes | relies on external nodes |
| Automation | envelope on some controls | minimal | full curves per parameter | full curves per parameter |
Reading the matrix
The edit-model row separates the four most sharply. Audacity applies many changes directly to a working copy and keeps the original as a second copy, which makes a destructive repair simple but a late change of mind expensive. Ocenaudio sits a step closer to a graph, offering cached undo over region edits. Ardour and Reaper both hold a full non-destructive graph, so a node can be reordered or bypassed long after the take was captured. For restoration work that means Audacity and Ocenaudio carry their own spectral repair, while Ardour and Reaper expect those jobs to arrive as external nodes.
The internal-format row is closer than it looks. All four keep a wide float format for the working signal, so none of them will betray a chain by rounding at an intermediate step. Reaper margins ahead only in the width it reserves before the output stage. What actually decides the sound of a finished master is almost never this row; it is the metering row and the gain discipline that follows from it.
Metering is where a delivery decision is made, and here the four diverge more usefully. Only two of the four present a loudness reading and a true-peak reading by default. A reader who works with the other two can still hit a loudness target, but must reach for an external meter to do it, and must then trust that meter's integration window. Routing depth follows a similar pattern: the two desks built around a graph offer buses, inserts and side-chain paths, while the two built around single-bus editing do not.
The desk keeps no affiliation with any of the four. Where a name appears, it is used to identify the object of comparison, and every number in the matrix describes engineering behaviour rather than quality or seniority.