Nøytral lyd, rett lyd og target frekvensrespons
Nøytral lyd, rett lyd og target frekvensrespons.
Her er det en del misforståelser ute og går.
Nå om dagen så lytter jeg til F105 i Rom2, og sammenligner direkte med F2. Frekvensresponsen for disse veldig forskjellige systemene er ganske lik, decay er mere forskjellig, og lyden er i hvertfall ikke lik.
Blir de likere om en justerer frekvensrespons til samme target? Nei. De vil alltid låte forskjellig, med forskjellig tonal balanse, fordi spredningen er forskjellig, og nyanser i lydkarakteren for horn og drivere spiller også inn. Selvsagt sett bort i fra kapasitet, men så lenge en ikke spiller for høyt, er det irrelevant.
Så lenge spredningen er forskjellig, er det ikke mulig å bruke eq for å få det til å låte likt. Decay - dvs hvordan lyden dør ut - blir da forskjellig, og det påvirker tonal balanse på transienter i forhold til kontinuerlige, jevne toner. Vanlig eq påvirker tonal balanse på frekvensrespons steady-state og tonal balanse på decay helt likt, og kan ikke endre noe på hvor fort lyden dør ut, dvs lengden på decay utover i tid.
Så er det dette med valg av target. Det er nå veldig populært at denne skal være med fallende respons oppover i frekvens, med et fall på rundt 10dB fra 20Hz til 20KHz, gjerne med referanse til et Bruel & Kjær paper ("Bruel & Kjær-kurven") fra 70-tallet(?) en gang, samt papers fra Harman som viser til lytteeksperimenter der det ble konstatert at folk flest foretrekker en slik fallende kurve.
Denne fallende kurven er rett kun når akustikken i rommet sammen med høyttalernes spredning er slik at det blir til en jevnt fallende kurve. Et rom med litt skjev balanse - typisk et rom med "akustikktiltak" i form av noen tynnere dempeplater her og der - samt en høyttaler som går fra rundstrålende i bassen til ganske snever spredning i toppen. Akkurat slik som det ser ut på bildene fra rommet med Harman-eksperimentene, og slik grafene fra høyttalerene er - jevn, flott spredning som tilter nedover som unnarennet i Vikersundbakken (Her kan det også stå "som nordvestrenna på Austabotntind", velg selv den analogien som faller best i smak.).
Med andre høyttalere og andre rom blir det ikke nødvendigvis slik. Det blir hverken nøytral eller "rett" lyd ved å innføre kutt i diskanten for å få til denne tilten, tonal balanse nedover i mellomtonen blir rar, og selv bassen blir ikke helt rett, fordi det gjerne låter bedre med en litt annen target for bass-boost.
Typisk så vil frekvensresponsen i et bra rom med jevnere demping frekvensmessig bli ganske flat, mens mere livlig akustikk vil gi noe som ligner mer på denne fallende kurven.
Så hvordan skal en da gå frem for å få rett kurve?
Sett inn en høyttaler med flat frekvensrepons on-axis og så bra og jevn spredning som mulig, og dette blir til av seg selv. Det er ikke nødvendig med noen eq, bortsett fra justeringer på target i bassen, som kan gjøres på bassystemet.
Problemet oppstår når en skal gjøre romkorreksjon med et av de skikkelige systemene, fx Audiolense. Da må du velge en target å justere mot, og denne skal da altså ikke nødvendigvis være som et unnarenn i en hoppbakke. Et utgangspunkt kan være å måle systemet uten eq først, og så bruke dette for å velge tilt på target.
Frequency response for F205 systems in 4 different rooms (1/1 smoothing to only show tilt) - barely no tilt, quite similar, even though the rooms are very different acoustically:
Some more data for you, to compare and visualize slope of the frequency response.
First, the Bruel&Kjær preference curve, for reference:
Now I do not remember exactly if it starts to slope to down at 20 or 100Hz, but the slope should be 3dB/octave, as shown here.
So, let't see if ANY real speaker-room combinations are like that.
Here is the F205 in 5 different rooms:
Now this makes no sense, so some smoothing so we can see the slope:
Most of the rooms (already shown in a previous graph..) are quite close to flat, while one actually comes closer to that B&K target.
2 of the rooms are fixed into proper acoustic performance. They are like this - FLAT across 100->top, with a small lift below 100Hz in the bass:
So maybe its the room - a more live room gives more tilt. Here is decay for one of the treated rooms - FLAT slope above 100Hz (first 0ms line is affected by rise time of 1ms so does not represent the frequency response correctly, thus more deviations with dips and holes):
Here is decay for one of the non-treated spaces, lots of reverb here, but almost NO TILT:
So it wasn't the amount of reverb in the room that caused the tilt. A decent speaker will have a nice, close-to-flat response in any room with reasonably neutral frequency response of decay . So, for the F205 to show a tilted sloping response, requires not only a room with insufficient damping, but also improper damping that results in a bad room with a tilted spectral decay.
May be it is the speaker. F205 has a much more linear and flat power response than the typical box speaker. Let's look at some boxes, and find out.
JBL4349, larger box-speaker, in a treated room, L R and sum (V80 bass system, with level adjusted for some lift in the bass range):
No tilt!
Tiny JBL (with T140 bass-system) in a treated room:
Still none if that excessive tilt, only s small increase at the low end, mostly a lift in the bass range.
May be a non-treated space will be different. Small Revel (with T6 bass-system) in a non-treated living room:
Some tilt, but not 10dB down, and the top end does not drop much.
If we examine decay, we can see more of what happens. The first room with those larger JBL, treated room:
We see room contributions after 20ms are so low in level they do not contribute to the overall frequency response. which explains the no-tilt response.
Tiny Revel in non-treated room:
We see the increase below 600Hz is caused by very early reflections in the room.
Decay for F205 in non-treated living room:
No tilt, and we actually see that decay is much more flat across the frequency range, and early reflections are attenuated more evenly. The speaker makes a difference.
So to get a slightly tilted response requires a defective speaker or a defective room, and to get something that comes close to the 3db-slope-target requires BOTH speakers and room to be defective.
What causes the tilted response?
Drop in level at high frequencies is due to high frequency loss in air, room contribution drops due to same, can be speakers are not toed in towards listening position and drops off off-axis.
In most of these room-speaker combinations the listening position is quite close to speakers, typically around 2.5m. Move toward say 4m, and there will be a small drop at high frequencies.
Then look at the 20ms curve. In all non-treated spaces it drops off at high frequencies, and that can partially be due to speaker radiation narrowing at the top of the frequency range. But it is also caused by the physical properties of sound propagation in air, when sound travels a longer distance, like after being reflected between several surfaces inside a room, the high frequencies will be damped due to loss in the air.
The drop at high frequencies is embedded into standards describing target frequency responses for studios, control rooms (look up EBU). It is suggested to not correct this into a flat measured response at the listening position.
Below say 1KHz the room contribution in a small room gets more and more a mess of resonances as frequency goes down. Boundaries gets acoustically closer because wavelengths become smaller compared to distances between sound speaker and boundaries, and this also applies to distance between listener and boundaries (walls, floor, ceiling). This contribution is not desired, and is a major cause for coloration, loss of clarity, compromised imaging. Speakers with better power response and full frequency range radiation control tend to be better, but the room will still affect those properties. This is the "small room sound".
As seen in the graphs for the good treated rooms compared to non-treated, this can be fixed, and the fixed rooms show no build-up of room contribution towards lower frequencies at mid and lower-mid frequencies. Going down into the bass-range, there is an increase in late energy, but this does not really compromise the sound, as long as resonances are removed using eq on the bass-system.
The increased level below 100Hz is deliberate, it is adjusted to give a small lift in the bass range, and both level and shape of the frequency response in the bass range affect how the bass sounds like. This is tailored with level and eq adjustments on the bass-system.