Published On: 11 Jun 2025Categories: Featured, Technical

Timing, Filtering & Perception: Why Digital Audio is Never Truly Neutral.

Is your DAC truly neutral?

Many believe a DAC is a transparent conduit, simply passing along what was encoded. But anyone who’s explored digital filters knows this isn’t the case.

The DAC industry — even at affordable levels — openly offers filter coloration as a customisable option. That should raise a fundamental question:

If these filters alter what you hear, what does it mean for “fidelity”?

 

The Underlying Problem
Every digital-to-analogue converter reconstructs the waveform using a filter — usually a steep linear-phase sinc filter or a variant like minimum phase, apodizing, or hybrid.

But all these filters trade off between:

  • Frequency accuracy (how flat is the frequency response?)
  • Time-domain fidelity (how clean is the impulse response? How much pre/post ringing?)
  • Phase behavior (is the phase linear or minimum?)

The AES paper “Audibility of Typical Digital Audio Filters” by Jackson, Stuart & Capp (2014) showed that the choice of DAC filter measurably affects impulse response and group delay, and that trained listeners could hear differences even between filters that measured “flat.”

So while frequency-domain plots may look identical, the temporal behaviour — how the signal unfolds in time — is not.

 

Why it Gets Worse in the Chain
Filters don’t just exist in your DAC. They’re in:

  • The original A/D converter
  • Any sample rate conversion stages
  • DSP effects
  • Your DAC output filter

 

Each stage adds a tiny bit of smearing or misalignment. Over a whole production chain, this accumulates.

Bob Stuart and Peter Craven addressed this in their paper “A Hierarchical Approach for Audio Capture, Archive and Distribution” (AES 2019), arguing that:

“While frequency and time are dual domains, many standard tools (FFT, etc.) use global time windows. But our perception is local and adaptive — the brain reacts to signal edges, not frequency plots.”

Their approach uses time-symmetric b-spline filters and a joint AD/DA model to minimise cumulative blur — aiming not for abstract fidelity, but perceptual realism.

 

So What’s the Takeaway?

  • No DAC is “just passing through” the signal.
    Its filter always alters time/frequency behaviour.
  • You are hearing the sum of many filter choices, starting from the ADC and ending at your DAC.
  • Modern digital audio is not lossless in perception — even if the bits say it is.
  • New approaches like spline-based filtering and joint optimisation (e.g. MQA, QRONO, FOQUS) don’t just compress audio, they try to undo accumulated smearing.

 

Final Thought
If you’re open-minded, try listening not for brightness, loudness or EQ, but for coherence, depth and timing. Music is not just sound — it’s movement in time.

Sometimes, the most truthful playback isn’t what measures best on an FFT.

Peter Veth

Categories

Latest News