- Joined
- Nov 14, 2020
- Messages
- 426
- Likes
- 596
Hi out there,
a few weeks ago, I had a visit from someone who claimed that his homemade R2R NOS DAC, based on a TDA1541A chip from the 80s, sounded better than my state-of-the-art RME ADI-2/4 Pro SE.
After a short listening session with A/B and same volume (I have a switch box), it was clear to me that the R2R DAC simply lacked resolution, which, among other things, led to a collapse of the spatial impression and the loss of the subtle details I love so much.
His argument was that he preferred the R2R DAC because it didn't become fatiguing during extended listening sessions.
Based on this experience, I started a discussion with ChatGPT today, the outcome of which is very insightful and will certainly provide material for further discussion here in the forum.
I hope you enjoy reading this as much as I did.
Best DrCWO
Prompt:
There are people that claim that R2R DAC sound better. But there is no objective measurement that proofs this claim. Any idea why they claim that?
ChatGPT:
Short answer: because “sounds better” is usually a subjective judgment, and R-2R DACs have different imperfections than modern delta-sigma DACs. Some listeners prefer those imperfections, even if they don’t show up as better performance on standard measurements.
Here’s the longer, more nuanced explanation.
1. What R-2R DACs do differently
An R-2R DAC uses a ladder of precision resistors to directly convert digital values into analog voltages. Delta-sigma DACs (the vast majority today) instead:
2. Measurements optimize for audibility thresholds, not preference
Modern delta-sigma DACs measure far beyond human hearing limits:
R-2R DACs usually measure worse:
3. Distortion character matters more than distortion amount
Not all distortion sounds the same.
4. Time-domain behavior & filters (a big subjective factor)
Many R-2R DACs use:
There is no universally accepted metric for “time-domain naturalness,” especially for complex music.
5. Expectation bias & system interaction (this matters more than people admit)
Other non-technical reasons:
“It interacts with my system and preferences in a pleasing way.”
That doesn’t mean they’re lying—just that the perception is contextual.
6. Why there’s no objective proof
Because:
7. The most honest conclusion
“R-2R DACs sound better to me”
That can be completely true—without contradicting measurements at all.
If you want, I can:
Prompt:
This sounds interesting "how blind tests usually change people’s opinions"
ChatGPT:
Great topic — this is where a lot of audiophile debates quietly fall apart
I’ll walk through what blind tests typically show, why opinions often change, and why this is uncomfortable for many enthusiasts.
1. What blind tests usually find (in practice)
When DACs are compared under proper blind conditions (level-matched, fast switching, no visual cues):
2. Why sighted listening is so persuasive
Sighted listening activates top-down perception:
Your brain doesn’t just hear — it interprets.
Known influences:
Blind tests remove this scaffolding.
3. The “loss of narrative” effect
Many audiophiles report something like:
“Under blind testing, everything sounds kind of the same and boring.”
That’s not a coincidence.
Part of the enjoyment comes from:
4. Why opinions often change after blind testing
Three common outcomes:
A. Increased skepticism
People realize:
Instead of:
“R-2R is more musical”
They say:
“I sometimes prefer this DAC in long listening sessions with bright headphones.”
That’s a much stronger and more honest claim.
C. Doubling down (this happens too)
Some react defensively:
5. Famous examples (without internet drama)
No blind test has ever demonstrated a general preference for R-2R DACs.
6. Why blind tests feel “wrong” to some listeners
Blind tests:
If a difference disappears under blind conditions, it is not robust.
It may still exist for that person, but it’s not a property of the DAC alone.
7. The uncomfortable implication
Blind tests suggest:
8. The balanced takeaway
Blind testing doesn’t say:
“You’re imagining everything.”
It says:
“Your perception is deeply context-dependent.”
And that’s true for all human senses, not just hearing.
If you want, next we can dig into:
Prompt:
This "long-term blind tests change anything" is interesting too
ChatGPT:
This is a really good follow-up, because “short ABX tests are stressful and unrealistic” is probably the strongest criticism people make of blind testing. So let’s look carefully at what happens when tests are made long-term and more “natural”.
1. What people mean by long-term blind testing
Typically:
2. What long-term blind tests actually show
When done properly, results tend to fall into three patterns:
Pattern 1: Initial confidence fades
People start with:
“I’m sure I’ll notice it over time.”
After days or weeks:
“They’re more similar than I expected.”
Pattern 2: Preferences don’t correlate with DAC identity
Listeners may consistently prefer one, but:
Pattern 3: The difference disappears entirely
Many report:
“I stopped thinking about the DAC and just listened to music.”
That’s actually the strongest evidence of transparency.
If a component doesn’t demand attention, it’s doing its job.
3. Why long-term tests still don’t rescue DAC differences
People often expect subtle effects to accumulate over time — but the opposite happens.
Human perception adapts
4. The “listening fatigue” claim
You’ll often hear:
“This DAC is less fatiguing over long sessions.”
Under blind long-term conditions:
5. What does survive long-term blind tests
Some differences do persist:
These are all measurable, first-order effects.
6. Why audiophile lore predicts the opposite
Audiophile culture assumes:
“Subtle improvements reveal themselves with time.”
Psychology shows:
“Expectation effects weaken with time when feedback is removed.”
Long-term blind testing removes reinforcement:
7. A fascinating paradox
Many people report:
“I enjoy my system more after blind testing.”
Why?
8. The key conclusion
Long-term blind testing:
They don’t.
9. The most honest framing
A defensible statement is:
“I like this DAC, I know why, and I don’t need it to be universally better.”
An indefensible one is:
“It’s obviously more natural if you listen long enough.”
Long-term blind tests are precisely what challenge that claim — and that’s why they’re rarely embraced.
If you want, we can go deeper into:
Prompt:
This is interesting "how to design your own long-term blind test at home"
ChatGPT:
Great — this is where things get practical and surprisingly eye-opening. I’ll give you a realistic, home-doable long-term blind test that avoids the usual pitfalls and doesn’t require lab gear.
The goal is not to “win” an argument, but to find out whether you can hear a robust difference.
1. The core principle (keep this in mind)
You are not testing:
“Can I hear any difference if I concentrate?”
You are testing:
“Does one DAC reliably make me enjoy music more over time when I don’t know which one it is?”
That framing changes everything.
2. What you need (minimal setup)
Required
3. Step 1: Level matching (critical)
This is the most important step.
A 0.3 dB difference is easily perceived as “better clarity”.
Do this before blinding.
4. Step 2: Blind the identity (properly)
Best method (simple and robust)
5. Step 3: Long-term listening phase
Each phase should last:
6. Step 4: Light journaling (very important)
Once per day (or every few sessions), write one or two sentences max:
Examples:
Avoid guessing which DAC it is.
You are tracking experience, not traits.
7. Step 5: Switch, repeat, and add a twist
After the first period:
The placebo round
8. Step 6: Reveal and analyze
Only after all rounds are done:
9. How to interpret outcomes (honestly)
Outcome A: Clear, consistent preference
This is rare — but valid.
This is the most common result.
Congratulations:
10. Common mistakes to avoid
11. Why this works better than ABX
ABX tests:
12. Final thought (important)
If you do this honestly, one of two things happens:
Very few audiophiles do this because it risks killing cherished beliefs — but for someone genuinely curious, it’s one of the most enlightening things you can do in audio.
a few weeks ago, I had a visit from someone who claimed that his homemade R2R NOS DAC, based on a TDA1541A chip from the 80s, sounded better than my state-of-the-art RME ADI-2/4 Pro SE.
After a short listening session with A/B and same volume (I have a switch box), it was clear to me that the R2R DAC simply lacked resolution, which, among other things, led to a collapse of the spatial impression and the loss of the subtle details I love so much.
His argument was that he preferred the R2R DAC because it didn't become fatiguing during extended listening sessions.
Based on this experience, I started a discussion with ChatGPT today, the outcome of which is very insightful and will certainly provide material for further discussion here in the forum.
I hope you enjoy reading this as much as I did.
Best DrCWO
Prompt:
There are people that claim that R2R DAC sound better. But there is no objective measurement that proofs this claim. Any idea why they claim that?
ChatGPT:
Short answer: because “sounds better” is usually a subjective judgment, and R-2R DACs have different imperfections than modern delta-sigma DACs. Some listeners prefer those imperfections, even if they don’t show up as better performance on standard measurements.
Here’s the longer, more nuanced explanation.
1. What R-2R DACs do differently
An R-2R DAC uses a ladder of precision resistors to directly convert digital values into analog voltages. Delta-sigma DACs (the vast majority today) instead:
- Oversample heavily
- Use noise shaping
- Push quantization noise far above the audible band
- Rely on digital filtering and reconstruction
2. Measurements optimize for audibility thresholds, not preference
Modern delta-sigma DACs measure far beyond human hearing limits:
- THD+N below −120 dB
- Near-perfect linearity
- Vanishing jitter
R-2R DACs usually measure worse:
- Higher distortion
- Worse linearity at very low levels
- More sensitivity to temperature and resistor tolerance
3. Distortion character matters more than distortion amount
Not all distortion sounds the same.
- Delta-sigma DACstend to produce:
- Very low overall distortion
- Higher-order artifacts (though extremely quiet)
- Pre/post-ringing from sharp digital filters
- R-2R DACstend to produce:
- More low-order harmonic distortion (2nd, 3rd)
- Very little noise shaping
- Often simpler or NOS (non-oversampling) filtering
- “Warm”
- “Dense”
- “Natural”
4. Time-domain behavior & filters (a big subjective factor)
Many R-2R DACs use:
- Minimal filtering
- NOS designs (no oversampling)
- No pre-ringing
- More ultrasonic images
- Slight HF roll-off
There is no universally accepted metric for “time-domain naturalness,” especially for complex music.
5. Expectation bias & system interaction (this matters more than people admit)
Other non-technical reasons:
- R-2R DACs are expensive → expectation bias
- Often paired with tube amps or Class-A gear
- Slight tonal coloration can compensate for:
- Bright headphones
- Lean speakers
- Aggressive recordings
“It interacts with my system and preferences in a pleasing way.”
That doesn’t mean they’re lying—just that the perception is contextual.
6. Why there’s no objective proof
Because:
- DACs already exceed audibility thresholds
- Human perception is nonlinear and preference-based
- Standard measurements focus on accuracy, not euphony
- Blind tests usually show no consistent preference once levels are matched
7. The most honest conclusion
- Delta-sigma DACs are more accurate
- R-2R DACs are more characterful
- Preference ≠ fidelity
- “Better” depends on what you value:
- Truth to the signal
- Or pleasure in listening
“R-2R DACs sound better to me”
That can be completely true—without contradicting measurements at all.
- Explain why NOS DACs are especially controversial
- Break down which measurements might correlate better with preference
- Or discuss how blind tests usually change people’s opinions
Prompt:
This sounds interesting "how blind tests usually change people’s opinions"
ChatGPT:
Great topic — this is where a lot of audiophile debates quietly fall apart
I’ll walk through what blind tests typically show, why opinions often change, and why this is uncomfortable for many enthusiasts.
1. What blind tests usually find (in practice)
When DACs are compared under proper blind conditions (level-matched, fast switching, no visual cues):
- Listeners struggle to reliably distinguish DACs
- Preferences often flip randomly between trials
- Confidence drops sharply (“I’m not sure anymore”)
- Measured differences stop predicting perceived differences
- One DAC is R-2R and the other delta-sigma
- One costs 10× more
- One has dramatically better published specs
- Small
- Inconsistent
- Listener-dependent
2. Why sighted listening is so persuasive
Sighted listening activates top-down perception:
Your brain doesn’t just hear — it interprets.
Known influences:
- Price and brand reputation
- Physical weight and build quality
- Marketing narratives (“discrete ladder”, “no digital glare”)
- Prior beliefs (“measurements don’t tell the whole story”)
Blind tests remove this scaffolding.
3. The “loss of narrative” effect
Many audiophiles report something like:
“Under blind testing, everything sounds kind of the same and boring.”
That’s not a coincidence.
Part of the enjoyment comes from:
- Focusing attention
- Actively searching for differences
- Believing you’re hearing deeper into the recording
- No identity for the sound
- No story to attach to the experience
- Less emotional investment
4. Why opinions often change after blind testing
Three common outcomes:
A. Increased skepticism
People realize:
- Their confidence was inflated
- Differences are smaller than imagined
- Many descriptors were post-hoc rationalizations
- Measurements
- Controlled testing
- “Good enough” DACs
Instead of:
“R-2R is more musical”
They say:
“I sometimes prefer this DAC in long listening sessions with bright headphones.”
That’s a much stronger and more honest claim.
C. Doubling down (this happens too)
Some react defensively:
- Reject blind testing entirely
- Claim it “kills musicality”
- Argue stress or switching invalidates results
5. Famous examples (without internet drama)
- ABX tests repeatedly show transparency once DACs reach competence
- Studio engineers routinely use inexpensive delta-sigma DACs without complaint
- When trained listeners can detect differences, effect sizes are tiny
No blind test has ever demonstrated a general preference for R-2R DACs.
6. Why blind tests feel “wrong” to some listeners
Blind tests:
- Are analytical, not relaxing
- Emphasize short-term memory
- Remove ritual and familiarity
If a difference disappears under blind conditions, it is not robust.
It may still exist for that person, but it’s not a property of the DAC alone.
7. The uncomfortable implication
Blind tests suggest:
- DAC design is largely a solved problem
- Speakers, headphones, and room dominate the sound
- “House sound” of DACs is mostly expectation + tiny coloration
- Expensive upgrades
- Personal expertise
- Community status
8. The balanced takeaway
Blind testing doesn’t say:
“You’re imagining everything.”
It says:
“Your perception is deeply context-dependent.”
And that’s true for all human senses, not just hearing.
- Why trained listeners still fail blind tests
- Whether long-term blind tests change anything
- How manufacturers exploit cognitive biases
- Or how to enjoy hi-fi without falling into self-deception
Prompt:
This "long-term blind tests change anything" is interesting too
ChatGPT:
This is a really good follow-up, because “short ABX tests are stressful and unrealistic” is probably the strongest criticism people make of blind testing. So let’s look carefully at what happens when tests are made long-term and more “natural”.
1. What people mean by long-term blind testing
Typically:
- You live with Device A or B for days or weeks
- You don’t know which is which
- Switching is hidden or randomized
- Levels are matched
- You keep notes or just observe preference over time
2. What long-term blind tests actually show
When done properly, results tend to fall into three patterns:
Pattern 1: Initial confidence fades
People start with:
“I’m sure I’ll notice it over time.”
After days or weeks:
- Differences become harder to articulate
- Preferences drift
- Confidence drops
“They’re more similar than I expected.”
Pattern 2: Preferences don’t correlate with DAC identity
Listeners may consistently prefer one, but:
- When identities are revealed, the preference often doesn’t line up with the “expected” DAC
- Sometimes the cheaper or “worse-measuring” unit is preferred
- Sometimes preference flips across weeks
Pattern 3: The difference disappears entirely
Many report:
“I stopped thinking about the DAC and just listened to music.”
That’s actually the strongest evidence of transparency.
If a component doesn’t demand attention, it’s doing its job.
3. Why long-term tests still don’t rescue DAC differences
People often expect subtle effects to accumulate over time — but the opposite happens.
Human perception adapts
- The brain quickly normalizes constant stimuli
- Small colorations stop being noticeable
- Only large, persistent deviations remain salient
- Room acoustics matter a lot
- Speakers matter a lot
- DACs usually don’t
4. The “listening fatigue” claim
You’ll often hear:
“This DAC is less fatiguing over long sessions.”
Under blind long-term conditions:
- Fatigue correlates more with recordings, headphones/speakers, and volume
- DAC identity rarely predicts fatigue
- Fatigue reports often disappear when identity is hidden
5. What does survive long-term blind tests
Some differences do persist:
- Output level mismatches
- Frequency response deviations
- Noise floor issues
- Channel imbalance
- Filter choices (especially NOS vs OS)
- Broken or poorly designed gear
These are all measurable, first-order effects.
6. Why audiophile lore predicts the opposite
Audiophile culture assumes:
“Subtle improvements reveal themselves with time.”
Psychology shows:
“Expectation effects weaken with time when feedback is removed.”
Long-term blind testing removes reinforcement:
- No visual confirmation
- No praise from others
- No confirmation bias loop
7. A fascinating paradox
Many people report:
“I enjoy my system more after blind testing.”
Why?
- Less gear anxiety
- Less chasing differences
- More focus on music
- More confidence in “good enough”
8. The key conclusion
Long-term blind testing:
- Does not reveal hidden DAC magic
- Does not validate R-2R superiority claims
- Does reduce overconfidence and folklore
- Does shift focus to meaningful variables
They don’t.
9. The most honest framing
A defensible statement is:
“I like this DAC, I know why, and I don’t need it to be universally better.”
An indefensible one is:
“It’s obviously more natural if you listen long enough.”
Long-term blind tests are precisely what challenge that claim — and that’s why they’re rarely embraced.
- Why trained listeners still fail over time
- Whether NOS DACs are a special case
- How ABX methodology can be improved
- Or how to design your own long-term blind test at home
Prompt:
This is interesting "how to design your own long-term blind test at home"
ChatGPT:
Great — this is where things get practical and surprisingly eye-opening. I’ll give you a realistic, home-doable long-term blind test that avoids the usual pitfalls and doesn’t require lab gear.
The goal is not to “win” an argument, but to find out whether you can hear a robust difference.
1. The core principle (keep this in mind)
You are not testing:
“Can I hear any difference if I concentrate?”
You are testing:
“Does one DAC reliably make me enjoy music more over time when I don’t know which one it is?”
That framing changes everything.
2. What you need (minimal setup)
Required
- Two DACs (e.g. R-2R vs delta-sigma)
- A way to switch DACs without seeing which is active
- Hardware input switch
- Or two identical inputs on your amp
- Identical cables (don’t overthink this)
- Notebook or notes app
- A friend/partner to randomize assignments
- An SPL meter or measurement mic (even phone apps help)
3. Step 1: Level matching (critical)
This is the most important step.
- Play a 1 kHz sine wave
- Measure output at your listening position
- Match levels within ±0.1 dB if possible
(±0.2 dB minimum)
A 0.3 dB difference is easily perceived as “better clarity”.
Do this before blinding.
4. Step 2: Blind the identity (properly)
Best method (simple and robust)
- Label DACs A and B
- Have someone else secretly assign:
- Week 1 → A
- Week 2 → B
- Don’t tell you which is which
- Randomize order (coin flip)
- Which DAC is R-2R
- Which costs more
- Which you expect to like
5. Step 3: Long-term listening phase
Each phase should last:
- At least 5–7 days
- Normal listening only
- No A/B switching during a phase
- Analytical listening
- Memory tricks
- Hyper-focus on artifacts
6. Step 4: Light journaling (very important)
Once per day (or every few sessions), write one or two sentences max:
Examples:
- “Listened for 2 hours, no fatigue.”
- “Felt slightly bored / distracted.”
- “Really enjoyed vocals tonight.”
Avoid guessing which DAC it is.
You are tracking experience, not traits.
7. Step 5: Switch, repeat, and add a twist
After the first period:
- Switch DACs (still blind)
- Repeat the same process
The placebo round
- Have your partner secretly not change anything
- Tell you they did
- New “differences”
- A sudden preference
8. Step 6: Reveal and analyze
Only after all rounds are done:
- Reveal which DAC was which
- Compare notes
- Consistent preference aligned with one DAC
- Repeated language patterns
- Preference surviving the placebo round
9. How to interpret outcomes (honestly)
Outcome A: Clear, consistent preference
This is rare — but valid.
- The DAC difference matters to you
- No need for universal claims
This is the most common result.
- Differences are not robust
- Sighted listening likely inflated them
Congratulations:
- You’ve reached transparency
- Stop worrying about DACs
10. Common mistakes to avoid
- Switching mid-track
- “Hunting” for differences
- Reading forums during the test
- Changing volume
- Using unfamiliar music
- Short test periods (“I’ll know in 10 minutes”)
11. Why this works better than ABX
ABX tests:
- Short-term
- Stressful
- Analytical
- Tests preference, not detection
- Mimics real listening
- Lets biases decay naturally
12. Final thought (important)
If you do this honestly, one of two things happens:
- You free yourself from gear anxiety
- Or you justify your preference with confidence
Very few audiophiles do this because it risks killing cherished beliefs — but for someone genuinely curious, it’s one of the most enlightening things you can do in audio.
Last edited:
