Are you able to go beyond posting nothing but an unhelpfui personal attack and say what you are talking about?
B.
Because waveguides do not have a flat response (i.e. -6 dB/oct), so crossovers for them are much harder to do right.
What we hear in a room above some lower frequency, say 500 Hz, is dominated by the direct response and the coherent phase of this signal (the phases of individual frequency components relative to each other, not to the source) can be quite audible. Phase for a steady state response have no relevance because we don't hear a steady-state signal the same way as a transient one. Your discussion of phase did not take any of this into account and as such was not very meaningful.
Unfortunately as others have posted - there is no "auto-tune / quick fix" I am aware of.
I haven't experimented with active crossovers. It's probably my next venture.
What I am hoping to do is:
1. model the "perfect" acoustic slope for a particular driver
2. Measure the actual "in box" response
3. Subtract one from the other to derive the electrical transfer function required
4. Convert this function into the for example a particular DSP's format
5. Import, set time delay offset, re-measure, listen and repeat / tweak as required
I haven't experimented with active crossovers. It's probably my next venture.
What I am hoping to do is:
1. model the "perfect" acoustic slope for a particular driver
2. Measure the actual "in box" response
3. Subtract one from the other to derive the electrical transfer function required
4. Convert this function into the for example a particular DSP's format
5. Import, set time delay offset, re-measure, listen and repeat / tweak as required
I can't recall last seeing a clearer presentation of an innocent dream of audio engineering theory. The steps (if they actually could be performed) might result in the desired response in an anechoic chamber. Of course, sound in an anechoic chamber is awful, if I recall my experience in the world's largest long ago.
Or you could simply skip steps 1, 3, 4, and half of 5.
B.
Without for a minute agreeing with your imaginative notions about coherent phase, my critique is that phase at your ear - and each separately - and even when a single source/driver is used, is very scrambled and gets more scrambled when you so much as breathe. And before it gets to your pre-amp, it was well-scrambled during recording relative to the musical instrument.
Whether or not it meets your theoretical requirements for good stereo (that is, dependency on phase coherence as you say, at least above 500 Hz where wavelengths are getting shorter), the scrambling of phase information will happen.
B.
This is simply not true for the reasons that I stated, but believe what you want.
" my critique is that phase at your ear - and each separately - and even when a single source/driver is used, is very scrambled and gets more scrambled when you so much as breathe. "
Hi Ben,
Are you saying that if I were to move a microphone about 5mm on the same axis and re-measure the impulse response, there will be a visible change between the 2 impulse responses or their gated FFT derived frequency/phase responses?
If yes, what would be the cause and do you have data to show this?
Peter
Hi Ben,
Are you saying that if I were to move a microphone about 5mm on the same axis and re-measure the impulse response, there will be a visible change between the 2 impulse responses or their gated FFT derived frequency/phase responses?
If yes, what would be the cause and do you have data to show this?
Peter
Ben your continued rebuttal of well established engineering principals is becoming tiring. Do what you like but please stop telling others that accepted, tried and true methods are a waste of time.
Your philosophy of speaker building has it's place and some do choose to go down that route and that's fine. BUT it is not the only path one can follow, and many would argue that for the majority it is not the correct path.
Tony.
I don't think DSP is a solution to audio.
It involves dissecting music into events, placing events into microscopic boxes.
it then multiply some boxes, make some smaller, it rearrange some boxes, it changes their order, the number, multiply the amount or diminish their amount in other boxes.
Multiply the box contents by some numbers or divide or use logs.
Then you are believing that this is music. All it proves is how bad in the history of humanity our audio systems are, how tasteless, lifeless.
It involves dissecting music into events, placing events into microscopic boxes.
it then multiply some boxes, make some smaller, it rearrange some boxes, it changes their order, the number, multiply the amount or diminish their amount in other boxes.
Multiply the box contents by some numbers or divide or use logs.
Then you are believing that this is music. All it proves is how bad in the history of humanity our audio systems are, how tasteless, lifeless.
Nah. The driver SPL curves will most likely be quite different in your box/baffle.
You also need to take into account the baffle step, which is not included in the manufacturer data. It's usually best to use a good baffle step simulator for this. Vituix CAD has one. I like Tolvan Edge (old but good).
WOW! This post is over-the-top. Sorry, but years of testing has shown none of this to be true.
OK, back to Geddes imaginative notions on coherent phase information reaching your ears. What I have to post in exact reply to Peter's question will not be new to anyone who has experience in audio testing.
I sent a 1kHz sq wave to near full-range ESL panel (6 cells per panel). If you know audio, the RTA will have the familiar look of a square wave. At 1kHz, not sure the sound has much character to talk about. But if you wanted to make the best possible square waves, my full-range ESLs are among the best you are likely to ever find to ever make square square waves unless you have your own ESL panels.
And if you know audio, the following two plots will also look like the usual square wave: All Scrambled and barely resembling sq waves and scrambled slightly differently because, as Peter asked, one side is roughly 8.5 inches and the other 9.5, on axis ("roughly" since you can't define the exact distance to a panel, a mic, etc). Should be self-evident which is 8.5 inches.
Well, good-bye to coherent phase information and all that loose prattle on this forum about tweaking phase in XOs.
Peter - what do you think?
B.
Attachments
Last edited:
Yeah, digital has become the reference and with DSP you can tailor the sound as you like.
Thanks Ben, however I didn't qualify with the environment for the steps nor whether I had applied gating for pseudo-anechoic measurements.
I plan to follow the method using both gated and "full room" effects.
Its good to know the method is not flawed to achieve a desired, consistent outcome. Whether someone agrees the outcome is useful is another matter.
Fine. Let's take your step 1:
"1. model the "perfect" acoustic slope for a particular driver"
I don't know what the slope might be for any driver derived from theory or in practice until tested. Do you or perhaps wintermute knows how to decide?
But I do know that you I would run each driver and see how it outputs, distorts, resonates, takes power, etc. And then using your mature judgment, decide where you need to cut if off and with what slope and in relation to the other drivers. For example, if you like soft dome tweeters you need canny judgment (not perfect theory) about how low you can go.
Seems to me, that exactly zero of this empirical procedure could be in any way described as "'perfect' acoustic slope".
... and that's why you might as well start right off with step 2 even if it violates the Infinitely True Rules of Science as understood by wintermute, and skip 3 and 4.
B.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.