50-70% of soundtrack content is in the center channel.
Although many soundtracks contain some music in the center, the most important function of a center channel is maximum clarity of dialog.
Male voice fundamentals average * only 130 * Hz (female 205 Hz).
To be capable of comfortably covering deeper voices, a ‘safe‘ crossover (for argument's sake, 4th order) to bass would be about 80 Hz.
This would mean a mid-bass rather than a mid, with Fs of 50 or less, and or a Zobel.
Do you agree with these facts, or the desirability of such a low crossover for a mid for the center speaker, to reproduce male voices better than a sub/bass driver?
Although many soundtracks contain some music in the center, the most important function of a center channel is maximum clarity of dialog.
Male voice fundamentals average * only 130 * Hz (female 205 Hz).
To be capable of comfortably covering deeper voices, a ‘safe‘ crossover (for argument's sake, 4th order) to bass would be about 80 Hz.
This would mean a mid-bass rather than a mid, with Fs of 50 or less, and or a Zobel.
Do you agree with these facts, or the desirability of such a low crossover for a mid for the center speaker, to reproduce male voices better than a sub/bass driver?
One of the most annoying things in the world, to me, is having the guy from Radio Shack tell me that bass sounds are non-directional, and then he tries to sell me a $200 HT setup with overglorified PC speakers (YUCK!) as satellites and an 8" woofer in a decent box.
While it's true that the REALLY BASSY sounds are non-directional, what they catergorize as "bass" really isn't--listen to the bass man in Rockapella, and hear that his voice goes way below consumer HT crossover frequencies--and then try to tell me that's non-directional! I could bounce him around the room with a decent setup and you could hear where it's coming from without a hitch.
(wow, that was quite a tangent.)
For my HT setup, I'm actually aiming at having full-range speakers (down to 45hz, up to 20k) and then a subwoofer to reach everything down to 20 hz or so. I don't believe bass becomes non-directional until about 30 hz, so that's a nice compromise, since I don't want 8" woofers in all of my satellites. This is just my personal preference; I bet if you did your crossing over at 80 hz it would be just fine, and maybe even higher than that... but please, don't go above 150. That's just insane.
While it's true that the REALLY BASSY sounds are non-directional, what they catergorize as "bass" really isn't--listen to the bass man in Rockapella, and hear that his voice goes way below consumer HT crossover frequencies--and then try to tell me that's non-directional! I could bounce him around the room with a decent setup and you could hear where it's coming from without a hitch.
(wow, that was quite a tangent.)
For my HT setup, I'm actually aiming at having full-range speakers (down to 45hz, up to 20k) and then a subwoofer to reach everything down to 20 hz or so. I don't believe bass becomes non-directional until about 30 hz, so that's a nice compromise, since I don't want 8" woofers in all of my satellites. This is just my personal preference; I bet if you did your crossing over at 80 hz it would be just fine, and maybe even higher than that... but please, don't go above 150. That's just insane.
True... But..
Yes, bass is non-directional but, because of distortion heard from the cone, the localization made by the listener increases. There are many ways to work around this problem, one for example, is to create a down-firing sub.
In addition, to your Radiosh$t experience, I agree, I agree, and I agree. Many times I have journeyed in after work and had a demo of 'wicked' speakers for one hundred bucks. Horn sounded like horn pluse RS cr#$py sound, mid-bass was garbbled and the bass was peaky in that there was an obvious blip in the frequency response. I told the guy that sounded decent for 100 dollars and walked away...after buying my beloved Sennies ofcourse.
Rick, where did you pick-up this research? Basically what I'm doing for my centre channel is to create it such that the response where voices are concerned are not affected by a cossover as much as possible. Unfortunately my mains are already made and the Audax drivers I am using have an increase in the response around 3800-4500Hz (rough estimate from what I remember), and will need to x-over a little differnt in what I have done for the mains. I hope to timbre match as much as possibl but, could it be possible to buy Peerless drivers or some FRs from Tangband and keep the same 'voice' as my mains?
Yes, bass is non-directional but, because of distortion heard from the cone, the localization made by the listener increases. There are many ways to work around this problem, one for example, is to create a down-firing sub.
In addition, to your Radiosh$t experience, I agree, I agree, and I agree. Many times I have journeyed in after work and had a demo of 'wicked' speakers for one hundred bucks. Horn sounded like horn pluse RS cr#$py sound, mid-bass was garbbled and the bass was peaky in that there was an obvious blip in the frequency response. I told the guy that sounded decent for 100 dollars and walked away...after buying my beloved Sennies ofcourse.
Rick, where did you pick-up this research? Basically what I'm doing for my centre channel is to create it such that the response where voices are concerned are not affected by a cossover as much as possible. Unfortunately my mains are already made and the Audax drivers I am using have an increase in the response around 3800-4500Hz (rough estimate from what I remember), and will need to x-over a little differnt in what I have done for the mains. I hope to timbre match as much as possibl but, could it be possible to buy Peerless drivers or some FRs from Tangband and keep the same 'voice' as my mains?
A month ago I searched on Google for things like:
male +voice +fundamental
and
voice +intelligibility
IIRC there was some very good material from eg
a national voice/ singing academy, one of the pro equaliser makers, and the Schiller Institute.
All said much the same thing - female voice fundamentals average 205 Hz, male * only 130 * Hz.
A lot lower than many of us think . .
male +voice +fundamental
and
voice +intelligibility
IIRC there was some very good material from eg
a national voice/ singing academy, one of the pro equaliser makers, and the Schiller Institute.
All said much the same thing - female voice fundamentals average 205 Hz, male * only 130 * Hz.
A lot lower than many of us think . .
center
I've made a centerspeaker just a few month ago.
Its made of the following units:
Monacor SPH-100KEP
Monacor DT-250
At first a had a premade crossover, but that didnt suit the speakers (I found out the hard way). It just didnt ound good enough.
Then I was advices to insert a 1.order network, because off linearity off the speakers and it wold enhance the speak.
One thing I learnt is the importance off the high sounds when we speak. It is the "Tsssssss" and "hssssssss" that gives the openness off the center.
Just my experience !
I've made a centerspeaker just a few month ago.
Its made of the following units:
Monacor SPH-100KEP
Monacor DT-250
At first a had a premade crossover, but that didnt suit the speakers (I found out the hard way). It just didnt ound good enough.
Then I was advices to insert a 1.order network, because off linearity off the speakers and it wold enhance the speak.
One thing I learnt is the importance off the high sounds when we speak. It is the "Tsssssss" and "hssssssss" that gives the openness off the center.
Just my experience !
I think crossing your subwoofer higher than 100 hz is insane, let alone 150. I just finished a center channel today using two 3" Tang Band full rangers and I am very happy with it. I think the voices sound very natural, without that raspy top end that you get with diolouge through an oversensitive tweeter.
Speaking of bass directionality, that is one of my pet peeves about the Bose systems, and all those other inferior Sub/Sat set-ups. When you are watching the Lion King and all the characters voices are coming out of the center channel, and then suddenly Mufasa bellows from under some table in the corner of your room. Just a slight loss of realism there. My little dual 3" canter channel has no problem handling most of mufasas voice.... At least enough to make your brain think that the rest of it is coming from the center channel, and not the subwoofer, where it's REALLY coming from. Keep those subs crossed over nice and low, folks.
Speaking of bass directionality, that is one of my pet peeves about the Bose systems, and all those other inferior Sub/Sat set-ups. When you are watching the Lion King and all the characters voices are coming out of the center channel, and then suddenly Mufasa bellows from under some table in the corner of your room. Just a slight loss of realism there. My little dual 3" canter channel has no problem handling most of mufasas voice.... At least enough to make your brain think that the rest of it is coming from the center channel, and not the subwoofer, where it's REALLY coming from. Keep those subs crossed over nice and low, folks.
What is your SPL levels desired
out of the center channel ?
/crazy thoughts
1. 80hz - 150hz playback from
small drivers, let say 6" midrange
is not very loud. If you expect
high spl levels, I would look into
8" - 10" drivers.
2. Typcally, midrange driver capable of
good midbass performance (lets say down to 80hz) will lack in midrange
top end ... (1k - 2k)
3. Typically, midrange driver capable
of good midrange performance
(let say up to 2k - 4k) will lack in
midbass low end (80hz).
*****************************
Compromise is always a problem,
trying to fine tune the system
design.
Do you want a good sounding
midrange operating from 200hz - 2khz
cleanly *or* a midrange driver
capably of midbass operating from
80hz - 1khz ? Pick your poison.
If you use PHL "midrange" drivers,
either 6.5" or 8", you may operate
those around 200hz if using duals
(MTM center channel) and using
a 24db slope.
The gain/loss is;
I gain better performance
between 1khz - 2khz, vs. a loss in
performance from 80hz - 200hz.
If you were to use the Seas Excel's,
then you may gain better performance on the low end, but
sacrifice the top end. Some people
think 2k is where this driver breaks
up, proably 1.5khz - 2khz crossover
is needed.
I've use a single PHL 2520 crossed
at 220hz with a 18db slope, the midrange performace is incredible
and this driver is able to handle
my mono 600w amplifier. Very
strong vocals, but the catch 22 is...
You need a high output tweeter that
can operate high pass 2khz, the
2520 sensitivity is 100db 1w/1m.
Fun stuff
out of the center channel ?
/crazy thoughts
1. 80hz - 150hz playback from
small drivers, let say 6" midrange
is not very loud. If you expect
high spl levels, I would look into
8" - 10" drivers.
2. Typcally, midrange driver capable of
good midbass performance (lets say down to 80hz) will lack in midrange
top end ... (1k - 2k)
3. Typically, midrange driver capable
of good midrange performance
(let say up to 2k - 4k) will lack in
midbass low end (80hz).
*****************************
Compromise is always a problem,
trying to fine tune the system
design.
Do you want a good sounding
midrange operating from 200hz - 2khz
cleanly *or* a midrange driver
capably of midbass operating from
80hz - 1khz ? Pick your poison.
If you use PHL "midrange" drivers,
either 6.5" or 8", you may operate
those around 200hz if using duals
(MTM center channel) and using
a 24db slope.
The gain/loss is;
I gain better performance
between 1khz - 2khz, vs. a loss in
performance from 80hz - 200hz.
If you were to use the Seas Excel's,
then you may gain better performance on the low end, but
sacrifice the top end. Some people
think 2k is where this driver breaks
up, proably 1.5khz - 2khz crossover
is needed.
I've use a single PHL 2520 crossed
at 220hz with a 18db slope, the midrange performace is incredible
and this driver is able to handle
my mono 600w amplifier. Very
strong vocals, but the catch 22 is...
You need a high output tweeter that
can operate high pass 2khz, the
2520 sensitivity is 100db 1w/1m.
Fun stuff
intelligibility of voice
> What is the desired SPL levels out of the center channel ?
Don't have a number in mind. Enough for reasonable levels in a room 22 ft* 17 ft * 10 ft.
As I'm aiming for "high" spl levels, rather than driver size, I'm focussing on best intelligibility of voice.
> Do you want a good midrange operating from 200hz - 2khz cleanly *or* a midrange driver
capably of midbass operating from 80hz - 1khz ?
For intelligibility of voice, the latter is probably more important.
The best appears to be Manger, sold by Andre (alas big $, but only use one).
http://www.manger-msw.com/en/inhalt.html
This driver is 93 db efficient, can cover from 170 Hz to beyond 20 Hz.
It's not a conventional driver, and while it can handle many octaves, it's not a normal "fullrange" driver.
Perhaps should use the Seas Excel, Triangle or PHL 10 inch to "fill in" between Peerless XLS 12 and the Manger.
I was going to go virtual centre channel, but how well does it work for mutiple people and thus off center listening?
> What is the desired SPL levels out of the center channel ?
Don't have a number in mind. Enough for reasonable levels in a room 22 ft* 17 ft * 10 ft.
As I'm aiming for "high" spl levels, rather than driver size, I'm focussing on best intelligibility of voice.
> Do you want a good midrange operating from 200hz - 2khz cleanly *or* a midrange driver
capably of midbass operating from 80hz - 1khz ?
For intelligibility of voice, the latter is probably more important.
The best appears to be Manger, sold by Andre (alas big $, but only use one).
http://www.manger-msw.com/en/inhalt.html
This driver is 93 db efficient, can cover from 170 Hz to beyond 20 Hz.
It's not a conventional driver, and while it can handle many octaves, it's not a normal "fullrange" driver.
Perhaps should use the Seas Excel, Triangle or PHL 10 inch to "fill in" between Peerless XLS 12 and the Manger.
I was going to go virtual centre channel, but how well does it work for mutiple people and thus off center listening?
I'd also recommend a fullrange driver. Saying that female vocals are centered around 250Hz could very well be (I don't really know for a fact if it is or isn't), but I do know that without the high frequency reproduction of a fullrange driver or tweeter female vocals won't sound their best. And as it's always better to avoid a cross over in the most critical frequencies (voice) a fullrange driver maybe aided and either of the two extremes by another driver would seem the way to go.
singing vs speaking
Thanks, this breaks it up by the opera 'classes'.
I wonder how the opera singing voice compares to the spoken, probably *much larger* in range.
Other sites (I saw a few months ago) had the averages I quoted earlier.
> without the high frequency reproduction of a fullrange driver or tweeter, female vocals won't sound their best.
And it's always better to avoid a cross over in the most critical frequencies (voice)
You're right there.
Thanks, this breaks it up by the opera 'classes'.
I wonder how the opera singing voice compares to the spoken, probably *much larger* in range.
Other sites (I saw a few months ago) had the averages I quoted earlier.
> without the high frequency reproduction of a fullrange driver or tweeter, female vocals won't sound their best.
And it's always better to avoid a cross over in the most critical frequencies (voice)
You're right there.
My Physics text
My physics textbook states that the average voice (dialogue) frequency for males is 120Hz. and for females, 205Hz (I think, can't really remember). I was thinking of designing a centre channel using the 3" tangbands but, when modelled in WinISD I found that the frequency range becomes linear around 150-180Hz. far too high for the male voices, (89dB. est.) and for the males I would be running about 82-83dB. for the male voices, is this sufficient?
Also, does anybody have some response graphs from a completed project using the 3" tangbands and/or what are your listening experiences with these drivers?
My physics textbook states that the average voice (dialogue) frequency for males is 120Hz. and for females, 205Hz (I think, can't really remember). I was thinking of designing a centre channel using the 3" tangbands but, when modelled in WinISD I found that the frequency range becomes linear around 150-180Hz. far too high for the male voices, (89dB. est.) and for the males I would be running about 82-83dB. for the male voices, is this sufficient?
Also, does anybody have some response graphs from a completed project using the 3" tangbands and/or what are your listening experiences with these drivers?
Intelligibility for voice
Male voices indeed have some low frequency content but I think most recording engineers will roll off a lot of the low - 75 hz would be low - before the sound ever gets to your HT - I'd measure it before going to all the trouble of making it flat to 50 hz. There has been a lot of theoretical work done of intelligibility and the bottom line is that the inteligibility of a sound in a room is the ratio of direct to reflected sound times a factor reflecting (pardon) the time domain of the reflections. Thus you control inteligibllity by controlling reflection, generally, not frequency response.
In the frequency department, Bell Labs did a ton of work on this subject regarding intelligibility and frequency and basically found that the 1000 hz range plus or minus an octive is the key to inteligibility for human voice. I would sum up by saying that excellent transients in the 100 to 4000 hz range with high directivity toward the user (ie no reflections) would be the design goal of a center speaker. The best way I know of to minimize reflection is to move the source close to the user, and absorb sound at spots likely to be causing reflections.
The other problem with Low end is because it is less directional as emmitted, it tends to be reflected more in a room and in small square rooms pure mud results. I'll shut up - but my advice is don't go to nuts with low end on the center channel - it could reduce what you are seeking to increase.
Male voices indeed have some low frequency content but I think most recording engineers will roll off a lot of the low - 75 hz would be low - before the sound ever gets to your HT - I'd measure it before going to all the trouble of making it flat to 50 hz. There has been a lot of theoretical work done of intelligibility and the bottom line is that the inteligibility of a sound in a room is the ratio of direct to reflected sound times a factor reflecting (pardon) the time domain of the reflections. Thus you control inteligibllity by controlling reflection, generally, not frequency response.
In the frequency department, Bell Labs did a ton of work on this subject regarding intelligibility and frequency and basically found that the 1000 hz range plus or minus an octive is the key to inteligibility for human voice. I would sum up by saying that excellent transients in the 100 to 4000 hz range with high directivity toward the user (ie no reflections) would be the design goal of a center speaker. The best way I know of to minimize reflection is to move the source close to the user, and absorb sound at spots likely to be causing reflections.
The other problem with Low end is because it is less directional as emmitted, it tends to be reflected more in a room and in small square rooms pure mud results. I'll shut up - but my advice is don't go to nuts with low end on the center channel - it could reduce what you are seeking to increase.
In the frequency department, Bell Labs did a ton of work on this subject regarding intelligibility and frequency and basically found that the 1000 hz range plus or minus an octive is the key to inteligibility for human voice. I would sum up by saying that excellent transients in the 100 to 4000 hz
Seems to reinforce my previous ignored babble
I'd rather have PHL's operating from 200hz to 2khz
as opposed to any other driver performing from
80hz - 1khz...
Simple test reveals all.
Listen to a woofer with an adjustable crossover,
perhaps 18db slope plus or minus 6db slope.....
Set the crossover to low pass 80hz, 100hz, 150hz
and 200hz. Listen to the midrnage performance.
There is much happening low pass 100hz when it comes
to midrange as opposed to doing the same test
using a great midrnange driver and low passing
it from 1khz - 3khz sweeps.. Do the test to determine
what you like.
My preference ->More is gained by choosing a midrange
driver capable of the upper band, 2khz - 3khz.
I agree with Andre, the center channel should be designed
the same as the two front mains (left + right speakers)
within reason of course, at least use the same design
methodology as the mains.
Centre Channel design
Yes, that was and is now my goal in centre channel design...To continue with the design principals as my mains were designed from. Main reason: Timbre matching. I'm usually very 'pushy' when it comes to desiging loudspeaker systems for friends where they want to use Pioneer rears and use Sony fronts...it just doesn't work.
I guess the main reason why I thought about using the Tangbands was the cost...38 bucks for drivers! No crossover! Ah well, guess I'll stick with the single 5.25" Audax and TW010E1 up front!
Yes, that was and is now my goal in centre channel design...To continue with the design principals as my mains were designed from. Main reason: Timbre matching. I'm usually very 'pushy' when it comes to desiging loudspeaker systems for friends where they want to use Pioneer rears and use Sony fronts...it just doesn't work.
I guess the main reason why I thought about using the Tangbands was the cost...38 bucks for drivers! No crossover! Ah well, guess I'll stick with the single 5.25" Audax and TW010E1 up front!
A fair summary?
TheoM very interesting. Your logic is a reason to move the speakers close/ treat the room, or use dipoles. Have you read?
http://www.linkwitzlab.com/reproduction.htm
I’m intending to build a Linkwitz based dipole, like his Phoenix, a loudspeaker system of the “highest sonic performance”. http://www.linkwitzlab.com/builtown.htm
. . . Many who’ve heard it agree:
http://www.madisound.com/cgi-bin/archive_discuss.cgi?read=219913
Your ideas make me think that for the center bass, I should build the Phoenix bass panels, but not the subs. I will still do subs for L & R.
BTW, I like his sub design too: the Thor
http://www.linkwitzlab.com/thor-intro.htm
Compact and deep
> The 1000 Hz range +/- an octave is the key to intelligibility for human voice.
That’s 500 - 2 kHz.
- Odd when fundamentals average either 130 or (female) 205 Hz.
So this misses the fundamentals ??
100 to 4000 Hz range? I would have thought 100 to just 1 or - 2 kHz.
But around 3 kHz hearing is most sensitive, so you’re probably right,
maybe 100 to 3 kHz.
Thylantyr
you think the two frequency ranges are mutually exclusive?
Especially PHLs can cover wide ranges. The PHL 3430 has a range up to 3 kHz.
It’s Fs is about 50 Hz, Andre runs it up to 2 kHz, and rates it very well.
The *original question* was: what best Hz for a crossover from mid to bass.
It looks like 100 Hz is a good target. Thylantar you’re right, a midbass operating from 100 Hz to say 1.5 kHz would then be best.
A very good tweeter could cover the rest up.
Say Seas Millennium, or better for dynamics and detail a Raven R3, or Esg1 . . .
TheoM very interesting. Your logic is a reason to move the speakers close/ treat the room, or use dipoles. Have you read?
http://www.linkwitzlab.com/reproduction.htm
I’m intending to build a Linkwitz based dipole, like his Phoenix, a loudspeaker system of the “highest sonic performance”. http://www.linkwitzlab.com/builtown.htm
. . . Many who’ve heard it agree:
http://www.madisound.com/cgi-bin/archive_discuss.cgi?read=219913
Your ideas make me think that for the center bass, I should build the Phoenix bass panels, but not the subs. I will still do subs for L & R.
BTW, I like his sub design too: the Thor
http://www.linkwitzlab.com/thor-intro.htm
Compact and deep
> The 1000 Hz range +/- an octave is the key to intelligibility for human voice.
That’s 500 - 2 kHz.
- Odd when fundamentals average either 130 or (female) 205 Hz.
So this misses the fundamentals ??
100 to 4000 Hz range? I would have thought 100 to just 1 or - 2 kHz.
But around 3 kHz hearing is most sensitive, so you’re probably right,
maybe 100 to 3 kHz.
Thylantyr
you think the two frequency ranges are mutually exclusive?
Especially PHLs can cover wide ranges. The PHL 3430 has a range up to 3 kHz.
It’s Fs is about 50 Hz, Andre runs it up to 2 kHz, and rates it very well.
The *original question* was: what best Hz for a crossover from mid to bass.
It looks like 100 Hz is a good target. Thylantar you’re right, a midbass operating from 100 Hz to say 1.5 kHz would then be best.
A very good tweeter could cover the rest up.
Say Seas Millennium, or better for dynamics and detail a Raven R3, or Esg1 . . .
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.