What about F6B or C6B vs Sierra V2 or Sierra LX?
It essentially shows a tie between F6B and C6B.
| Metric | Better direction | C6B | F6B | Advantage |
|---|
| NBD_ON (smaller is better) | ↓ | 0.300 | 0.271 | F6B |
| NBD_LW (smaller is better) | ↓ | 0.291 | 0.264 | F6B |
| NBD_SP (smaller is better) | ↓ | 0.256 | 0.263 | C6B |
| NBD_PIR (smaller is better) | ↓ | 0.240 | 0.242 | C6B |
| SM_PIR (larger is better) | ↑ | 0.939 | 0.944 | F6B |
| SM_SP (larger is better) | ↑ | 0.948 | 0.954 | F6B |
| LFX (smaller is better) | ↓ | 39.6 Hz | 41.0 Hz | C6B |
| AAD_ON (smaller is better) | ↓ | 0.684 dB | 0.465 dB | F6B |
| Olive Preference Score (larger is better) | ↑ | 6.52 | 6.53 | essentially tied |
| Preference + subwoofer (larger is better) | ↑ | 8.40 | 8.47 | F6B |
The important point was that the
raw Olive scores were virtually identical: 6.52 vs. 6.53. The F6B had the cleaner on-axis response by NBD/AAD measures, while the C6B had slightly better bass extension and marginally better PIR NBD.
The four variables actually used in the original Olive regression were
NBD_ON, NBD_PIR, LFX, and SM_PIR. Their approximate relative weights in the paper were
31.5%, 20.5%, 30.5%, and 17.5%, respectively.
And here is the comparison to Ascend Sierra-1 V2:
| Metric | Better direction | C6B | Ascend Sierra-1 V2 | Advantage |
|---|
| NBD_ON (smaller is better) | ↓ | 0.300 | 0.484 | C6B |
| NBD_LW (smaller is better) | ↓ | 0.291 | 0.324 | C6B |
| NBD_SP (smaller is better) | ↓ | 0.256 | 0.269 | C6B |
| NBD_PIR (smaller is better) | ↓ | 0.240 | 0.251 | C6B |
| SM_PIR (larger is better) | ↑ | 0.939 | 0.944 | Sierra-1 V2 |
| SM_SP (larger is better) | ↑ | 0.948 | 0.969 | Sierra-1 V2 |
| LFX (smaller is better) | ↓ | 39.6 Hz | 42.5 Hz | C6B |
| AAD_ON (smaller is better) | ↓ | 0.684 dB | 0.321 dB | Sierra-1 V2 |
| Olive Preference Score (larger is better) | ↑ | 6.52 | 5.91 | C6B |
| Preference + subwoofer (larger is better) | ↑ | 8.40 | 7.92 | C6B |
The analysis is using way too much precision. I will need to ask it to fix that but for now, there it is.
Commentary from AI on above:
"The difference is about:
6.52−5.91=0.61
preference points.
Using the unchanged Olive equation, the Sierra loses primarily because of
NBD_ON:
| Component | Sierra-1 V2 relative to C6B |
|---|
| NBD_ON | −0.46 |
| NBD_PIR | −0.03 |
| LFX | −0.13 |
| SM_PIR | +0.01 |
| Net | ≈ −0.61 |
So the interesting result is that the two speakers are actually quite close in their
spatially averaged behavior. Their PIR NBD differs by only about 0.010, and the Sierra actually has slightly higher PIR and sound-power smoothness.
The
main penalty for the Sierra-1 V2 is its on-axis NBD:
- C6B: 0.300
- Sierra-1 V2: 0.484
That one term accounts for roughly
three quarters of the total score difference.
Also noteworthy is AAD_ON: by that metric the
Sierra is substantially better, 0.321 versus 0.684 dB. This is possible because AAD and NBD are measuring different things: AAD reflects broad deviation from the 200–400 Hz reference level, whereas NBD penalizes local variation within ½-octave bands."