One:
If the Left, Center, and Right speakers are not time, phase, and FR matched, you get really weird colorations - not just from lateral pans across the soundstage, but strange, hollow sound on full-stage symphonic tracks. The only way to get get matching in all these domains is to use the same drivers and crossovers in the frequency range mentioned above - at least the range from 300 Hz to 5 kHz or so. Phase/amplitude mismatches are more tolerable at higher and lower frequencies, but not in the midband.
Although it is conventional practice to digitally auto-EQ dissimilar speakers, this doesn't work very well - partly because EQ-ing in the frequency domain can actually make time and phase errors worse, making the Center speaker even more dissimilar than the Left and Right speakers. I would also avoid side-by-side MTM layouts - these are inherently colored, no matter how flat they measure (the ears are extremely sensitive to lateral lobing and translate it into spectral coloration).
Two:
It seems counterintuitive, but improvements in CSD/time response can sound quite similar to improvements in IM distortion. I've been designing speakers a long time, and I have trouble telling these two parameters apart. I don't really know why a speaker sounds "clearer" or more open until I measure.
Three:
Many people, including industry professionals, are fooled by peaks in the 2~5 kHz region into thinking the peaky speaker is "sharper" or more focussed-sounding. JBL intentionally made the L100 "Century" have the same peaks as the Altec 604 Duplex for just this reason - JBL isn't a stupid company, and the midrange peaks were intentional, since that's what their market wanted. Midrange peaks, punchy bass, and wide dynamic range defined the "West Coast Sound" for many years. In the past decade, frequency response has gotten flatter, particularly after self-powered internally equalized studio monitors became the industry standard.
This "punchy" West Coast sound is also the de facto standard for movie equalization and theater sound - Altec and JBL were headquartered in Los Angeles, movie capital of the world, and both firms were created by the movie-sound division of Western Electric. West Coast Sound is movie sound.
As a result, it's no surprise movie dialog (mixed on Altecs or JBL's) can sound murky on a conventional audiophile speaker balanced for the most natural-sounding music playback. One way to resolve the tension between midrange-peaky HT speakers and speakers balanced for music is selecting drivers with the lowest possible IM distortion and cleanest CSD's - that'll be good for voice intelligibility and music.
If the Left, Center, and Right speakers are not time, phase, and FR matched, you get really weird colorations - not just from lateral pans across the soundstage, but strange, hollow sound on full-stage symphonic tracks. The only way to get get matching in all these domains is to use the same drivers and crossovers in the frequency range mentioned above - at least the range from 300 Hz to 5 kHz or so. Phase/amplitude mismatches are more tolerable at higher and lower frequencies, but not in the midband.
Although it is conventional practice to digitally auto-EQ dissimilar speakers, this doesn't work very well - partly because EQ-ing in the frequency domain can actually make time and phase errors worse, making the Center speaker even more dissimilar than the Left and Right speakers. I would also avoid side-by-side MTM layouts - these are inherently colored, no matter how flat they measure (the ears are extremely sensitive to lateral lobing and translate it into spectral coloration).
Two:
It seems counterintuitive, but improvements in CSD/time response can sound quite similar to improvements in IM distortion. I've been designing speakers a long time, and I have trouble telling these two parameters apart. I don't really know why a speaker sounds "clearer" or more open until I measure.
Three:
Many people, including industry professionals, are fooled by peaks in the 2~5 kHz region into thinking the peaky speaker is "sharper" or more focussed-sounding. JBL intentionally made the L100 "Century" have the same peaks as the Altec 604 Duplex for just this reason - JBL isn't a stupid company, and the midrange peaks were intentional, since that's what their market wanted. Midrange peaks, punchy bass, and wide dynamic range defined the "West Coast Sound" for many years. In the past decade, frequency response has gotten flatter, particularly after self-powered internally equalized studio monitors became the industry standard.
This "punchy" West Coast sound is also the de facto standard for movie equalization and theater sound - Altec and JBL were headquartered in Los Angeles, movie capital of the world, and both firms were created by the movie-sound division of Western Electric. West Coast Sound is movie sound.
As a result, it's no surprise movie dialog (mixed on Altecs or JBL's) can sound murky on a conventional audiophile speaker balanced for the most natural-sounding music playback. One way to resolve the tension between midrange-peaky HT speakers and speakers balanced for music is selecting drivers with the lowest possible IM distortion and cleanest CSD's - that'll be good for voice intelligibility and music.
Re: Re: Clarity of voices ~ horses for courses
Hi,
Is your memory that good or that accurate ? I cannot speak for your
answering machine but you would have to go back a long time to
be able to make a pure analogue telephone call.
Telephony voice quality I can safely say is one of my expert areas.
/sreten.
(Senior Technical Consultant - Telecoms)
planet10 said:
Hi,
Is your memory that good or that accurate ? I cannot speak for your
answering machine but you would have to go back a long time to
be able to make a pure analogue telephone call.
Telephony voice quality I can safely say is one of my expert areas.
/sreten.(Senior Technical Consultant - Telecoms)
I've had an question about vocal clarity that has been bugging me for more than a dozen years and this might be a good thread to raise it in.
In the house I had the following set up. The speakers were KEF B110 with a Celestion HF1300 for a tweeter. The amp was a reasonable Technics and the tape deck a Nakamichi (sorry can't spell). But I found that when I played the same cassettes in the car I could always hear the vocals more clearly. Now the car was a Peugeot 504 wagon and the system was a very basic radio/cassette combination (no Dolby) playing through 4 cheap full range speakers. Two were in the rear section and the front two were aimed at my shins. Now for the life of me I couldn't see why I could hear the lyrics more clearly in the car than in the house but that was consistently the case. I mentioned this to a friend who had a music degree and whose sister was in a band and she noticed the same thing. She said she always "went for a drive" when she wnated to hear the words.
Some have said that because the car system had no bass or treble it was putting all the enrgy into the midrange. But I also saw an article in Electronics World that said that in another area (non audio) engineers were deliberatly introducing "noise" to improve the extraction of the wanted signal. I gather the noise spectrum for most cars is very heavily weighted towards low frequency end. Is there some interesting mechanism working here. I have even wondered if we have evolved to hear voices more clearly in the face of other noise pollution, battles, high winds (or earthquakes!).
Any ideas or research into this?
In the house I had the following set up. The speakers were KEF B110 with a Celestion HF1300 for a tweeter. The amp was a reasonable Technics and the tape deck a Nakamichi (sorry can't spell). But I found that when I played the same cassettes in the car I could always hear the vocals more clearly. Now the car was a Peugeot 504 wagon and the system was a very basic radio/cassette combination (no Dolby) playing through 4 cheap full range speakers. Two were in the rear section and the front two were aimed at my shins. Now for the life of me I couldn't see why I could hear the lyrics more clearly in the car than in the house but that was consistently the case. I mentioned this to a friend who had a music degree and whose sister was in a band and she noticed the same thing. She said she always "went for a drive" when she wnated to hear the words.
Some have said that because the car system had no bass or treble it was putting all the enrgy into the midrange. But I also saw an article in Electronics World that said that in another area (non audio) engineers were deliberatly introducing "noise" to improve the extraction of the wanted signal. I gather the noise spectrum for most cars is very heavily weighted towards low frequency end. Is there some interesting mechanism working here. I have even wondered if we have evolved to hear voices more clearly in the face of other noise pollution, battles, high winds (or earthquakes!).
Any ideas or research into this?
Jonathan Bright said:
There's a large part of your answer.... playing Dolby B tapes back on non Dolby equipment will give you a big hump in the mids as the NR circuit is not complete without the playback Dolby B. Add a car system that rolls off at each end and it's a mid fest.
Nakamichi's (had 2) were strange beasts as the tapes made on them sounded great when played back on the Nakamichi, but play back on some other tape deck could yield very different and horrible results.
Re: Re: Re: Re: Clarity of voices ~ horses for courses
Well for a local call on a pure analogue switch it would have had
some intrinsic bass filtering but none at all at the high end other
the the physical characteristics.
The rigorous requirements for band limiting Telephony are for the
same reasons as CD, aliasing distortion, and this bandlimiting is
the major limitation in the percieved quality of Telephony calls
given all other factors are optimum.
To go back to the original post,
The low end depends on male/female/ child/ language.
The top end is constant and corresponds to the peak in the ears
sensitivity found in loudness curves, not usually shown like this :
The point i'm trying to make is 3.5KHz is only an adequate bandwidth.
The criticality of the range is reducing either end causes unnacceptable
quality loss, it is not the case that only these frequencies are needed.
All high quality codec standards start off by extending the bandwidth
at both ends, typically (~150) to 8kHz.
/sreten.
planet10 said:
Well for a local call on a pure analogue switch it would have had
some intrinsic bass filtering but none at all at the high end other
the the physical characteristics.
The rigorous requirements for band limiting Telephony are for the
same reasons as CD, aliasing distortion, and this bandlimiting is
the major limitation in the percieved quality of Telephony calls
given all other factors are optimum.
To go back to the original post,
The low end depends on male/female/ child/ language.
The top end is constant and corresponds to the peak in the ears
sensitivity found in loudness curves, not usually shown like this :
An externally hosted image should be here but it was not working when we last tested it.
The point i'm trying to make is 3.5KHz is only an adequate bandwidth.
The criticality of the range is reducing either end causes unnacceptable
quality loss, it is not the case that only these frequencies are needed.
All high quality codec standards start off by extending the bandwidth
at both ends, typically (~150) to 8kHz.
/sreten.
I use an Audio-technica AT2020 to record voice-over for video and it is about -3db at 100. I find that voices are weak if I don't compensate for this, but anything lower than 100hz doesn't affect the tone of the voice. I usually boost 3 db here and then roll it off at 80hz to kill rumble. I have also read that the female voice can have harmonics up to 12khz, and I have found that you kill some of the air if you roll off before this point. So I would say that 100 hz-12khz is actually the critical vocal range.
Lynn,
That’s very interesting about the "West Coast Sound".
Btw, lately for the first time in my life I’ve frequently played two tracks from the Eagles’ Hotel California. I think that album (amongst others, but notably that one) was engineered with a variant of WC/ rock sound with warm rich bass and boosted range around the cymbals, yet overall “smooth”.
The clearest sound by far that I ever heard coming out of a TV set was from a Bang & Olufson that seemed to have a midrange peak, without going overboard or sounding harsh.
To counteract dialogue sounding murky on speakers balanced for natural music playback, apart from using drivers with the lowest IM distortion and cleanest CSDs, what to do – use a tape loop switchable in/ out with an EQ, for some boost 1500 – 3500 Hz?
That’s very interesting about the "West Coast Sound".
Btw, lately for the first time in my life I’ve frequently played two tracks from the Eagles’ Hotel California. I think that album (amongst others, but notably that one) was engineered with a variant of WC/ rock sound with warm rich bass and boosted range around the cymbals, yet overall “smooth”.
The clearest sound by far that I ever heard coming out of a TV set was from a Bang & Olufson that seemed to have a midrange peak, without going overboard or sounding harsh.
To counteract dialogue sounding murky on speakers balanced for natural music playback, apart from using drivers with the lowest IM distortion and cleanest CSDs, what to do – use a tape loop switchable in/ out with an EQ, for some boost 1500 – 3500 Hz?
otto88 said:
That is an interesting question. Tonal balance between treated and untreated is pretty much the same, so they'd not suffer that way. AJ is gonna jump on me, but i don't know how long you'd be able to hold out on not EnABLing the LR after being exposed to the centre.
The FR125 is an excellent driver for use with HT receivers. And they do really well (to my mind best) in a small aperiodic box so typically high WAF.
dave
"i don't know how long you'd be able to hold out on not EnABLing the LR after being exposed to the centre."
Will you be doing kits to EnABL any drivers; if so how much skill is needed to apply properly, how long does it take, and how much might they likely be?
"The FR125 is an excellent driver for use with HT receivers . . so typically high WAF"
That's why I named it.
"And they do really well (to my mind best) in a small aperiodic box"
That raises a very good question -
http://www.diyaudio.com/forums/showthread.php?s=&threadid=115810
Thanks
Will you be doing kits to EnABL any drivers; if so how much skill is needed to apply properly, how long does it take, and how much might they likely be?
"The FR125 is an excellent driver for use with HT receivers . . so typically high WAF"
That's why I named it.
"And they do really well (to my mind best) in a small aperiodic box"
That raises a very good question -
http://www.diyaudio.com/forums/showthread.php?s=&threadid=115810
Thanks
otto88 said:
Yes. Until such time as i can get all the bits wholesale i have been gathering together kits from the local retail stores on an individual request basis. I am just now putting a kit to together for Mike in north central Oz... once i've tallied all the bits i'll have a solid number. Something like $30 + post for a basic set of nibs, holder, 1 bottle of flat, 1 of gloss. I'm also going to find some small bottles of puzzlecoat (optional) for drivers that benefit from pre-coat. Maybe a brush for applying the gloss. The guys in the UK have pretty much got local sourcing down, i expect that a kit could be assembled in Oz too.
dave
a late comment in reference to the original post:
Bell labs found that for reproduction to have a natural sound, the product of the upper and lower -3dB points should be 400,000 - hence the nominal 200 - 2000Hz of the earlier telephones.
Yes, I know that earlier telephones did not sound all that natural - except when the extremely narrow bandwidth is considered.
This has interesting implications for home hi fi. I recall one DIY project being published in the British press in the mid 70s which was a variable width bandpass device which adhered to this formula.
Bell labs found that for reproduction to have a natural sound, the product of the upper and lower -3dB points should be 400,000 - hence the nominal 200 - 2000Hz of the earlier telephones.
Yes, I know that earlier telephones did not sound all that natural - except when the extremely narrow bandwidth is considered.
This has interesting implications for home hi fi. I recall one DIY project being published in the British press in the mid 70s which was a variable width bandpass device which adhered to this formula.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.