TL;DR
Get audio and creator gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
An independent developer documented a method for recreating the behavior of a commercial spatial audio effect on Linux. Rather than identify the effect’s internal algorithms, the developer recorded its responses to carefully designed test signals, including impulses and sine sweeps. The source does not identify the commercial product or establish whether the resulting Linux effect matches it.
A developer has described how they set out to recreate a commercial spatial audio effect for Linux by recording and analyzing how the original program processed test signals. The account, published on the Inter res et verba blog, focuses on measuring the effect’s behavior rather than discovering its internal code or exact design.
The author wanted a Linux program that could process audio in real time and make it feel farther from the ears and less “on the ear”, while retaining the relevant information in the sound. They built on PipeWire, Linux’s audio system, and say they began with little audio experience. Early experiments used standard filters and psychoacoustic techniques, with an LLM serving as a sounding board.
The author then explored head-related transfer functions (HRTFs), which describe how sound from a direction reaches a listener’s ears. They found the MIT KEMAR database unsuitable for the desired effect because its measurements came from an anechoic chamber without room echoes. The SADIE II database, which includes room responses, sounded closer to the commercial effect they remembered, according to the account.
Rather than continue guessing at techniques such as “perceptual envelopment” and “late-field decorrelation,” the developer shifted to measuring the reference program. They generated test audio on Linux, played it through the commercial software on a friend’s laptop, and recorded the output. The test file included impulses at different levels and in different stereo configurations, with silence between signals so each response could be captured separately.
Measuring the Effect Instead of Guessing
The account offers a practical example of black-box audio analysis: a developer can learn about a program’s behavior from its inputs and outputs without access to its source code or design. That approach matters to Linux users seeking alternatives to commercial audio tools, and to audio developers who need a testable way to compare an implementation with a reference.
The method also places limits on what can be concluded. The article explains how to capture measurable properties, including decay and phase relationships, but the supplied material does not establish that the Linux project fully reproduces the commercial effect. It describes an investigative approach, not a verified product comparison or a claim of identical sound.
As an affiliate, we earn on qualifying purchases.
From HRTF Experiments to Test Signals
The developer’s first approach relied on choosing filters and spatial-audio techniques, then judging the result by ear. The HRTF experiments suggested that room response was part of the sound they wanted: an anechoic measurement sounded inward and “cheap” to them, while SADIE II’s room responses seemed closer to the reference.
Practical access to that reference was limited. The author says they could only use the commercial program on a friend’s laptop, making repeated, rapid comparisons difficult. Reading material from Signalsmith and audio engineers helped them identify features to measure, leading to a new plan: record a set of inputs once and compare their outputs.
An impulse—a signal with a single nonzero sample—can reveal a filter’s response over time. For a linear, time-invariant system, the output can be modeled as the input convolved with that response. In stereo, the author represented the system as four responses: left input to left and right output, and right input to left and right output. This model depends on the program behaving linearly and consistently over time.
The author also used sine sweeps, which distribute more energy across frequencies than a one-sample impulse and can help distinguish a response from recording noise. The account describes estimating the frequency response by dividing output by input, then using regularization to limit unreliable results where the input carries little energy. The source excerpt ends while discussing the sweep, so it does not report the full measurement results.
“I could study the behaviour of the filter itself.”
— The developer, writing on the Inter res et verba blog
As an affiliate, we earn on qualifying purchases.
What the Measurements Cannot Establish
The supplied account does not name the commercial program, say whether its processing is linear and time-invariant across all settings, or provide a completed comparison between the reference and the Linux implementation. It also does not state whether the developer ultimately released a finished tool. The author’s description of the effect and of the HRTF databases reflects personal listening judgments; the excerpt gives no independent listening test or objective match score.
The source ends during its discussion of sine-sweep measurement, after introducing a regularized estimate of the frequency response. Further results, including what the measurements showed and whether the method captured any processing that varies with input level, are not included in the provided material.
head-related transfer function HRTF headphones
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Further Results Would Show the Match
The next evidence needed would be the rest of the developer’s measurement results: how the impulse and sweep recordings compared, whether the reference passed the linearity checks, and what changes followed in the Linux implementation. A released build or a documented listening comparison could help establish how closely the recreation matches the commercial effect. The source material provided here does not say when, or whether, those details will appear.
microphone for recording impulse responses
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did the developer set out to recreate?
A real-time spatial audio effect that, in the author’s description, made sound feel farther away and less close to the ears. The commercial program is not named in the supplied source.
How did the developer study the commercial software?
They generated test audio on Linux, played it through the program on a friend’s laptop, and recorded the processed output. Their tests used impulses and sine sweeps to measure the system’s responses.
Why did the author try SADIE II after MIT KEMAR?
The author says the MIT KEMAR measurements were made in an anechoic chamber, without room echoes they wanted for the effect. SADIE II includes room responses, which they felt sounded closer to the remembered reference.
Does the account prove the Linux effect sounds identical?
No. The source describes a measurement method and early experiments, but the material provided does not report an independent comparison or establish that the recreation matches the commercial program.
Source: hn
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
