Clarity of voices ~ horses for courses

otto88

Member
2007-12-05 10:39 am
In the 40s Bell Labs' research into speech intelligibility established the critical range is (I think it was) 500 – 3500 Hz. Though telephone applications go down to about 200 hz.

Lower distortion is of course always better, but I was reading the HiFi Critic last night (editor Martin Colloms, the only audio mag worth reading) where re MP3s, comments were made that music warrants (of course) high bit rates or WAV files, but for voices - lower bit rates are ok.

I haven't had a HT system. Deep bass and explosive HT effects aside, music reproduction is demanding of a good system than home theatre. Many systems are used for both music and HT, but for me this isn't relevant, for a system to be used in a different room, only for home theatre. Few action movies, mostly drama and comedy.

Selecting tweeters (or mids) for this - how important is it to get the lowest distortion?
Maybe difficult to answer that question quantitatively, but the view expressed in HiFi Critic makes me think - not that much.

The pragmatic/ Devil's advocate view:
Admittedly on the telephone there is usually only one voice coming down the line, and the handset is next to our ears, but we can mostly hear the other person's voice fine, via a single (50c?) phone speaker.

So (as opposed to a full orchestra in flight, at the other end of the spectrum of audio demands) would eg $30-40 Vifa/ Seas level quality tweeters (and similar mids and bass) be more than up to the task of listening to drama at normal TV levels?

eg the Seas H1212 tweeter . .

Horses for courses?
 
otto88 said:
Lower distortion is of course always better, but I was reading the HiFi Critic last night (editor Martin Colloms, the only audio mag worth reading) where re MP3s, comments were made that music warrants (of course) high bit rates or WAV files, but for voices - lower bit rates are ok.

I don't know about that. Intelligibility of telephone (& the digital answering machine) is significantly less than when it was analog.

Selecting tweeters (or mids) for this - how important is it to get the lowest distortion?

Do keep in mind that just like amplifiers we only have test kit that really measures the grossest of distortions.

dave
 
Very few driver manufactures show the distortion graphs for their drivers and when they do they show them at little power. Why? Because most of them have high distortion.

There are low distortion drivers out but they are expensive.

Many of the people here don't use them and some actually paint little dots on crappy drivers to fool themselves (and others) into believing they have better drivers. :D Some will buy expensive speaker wire and claim they improve the sound - it's basic BS-

Bell Labs was right, use the lowest distortion drivers you can afford.
 

otto88

Member
2007-12-05 10:39 am
> Intelligibility of telephone (& the digital answering machine) is significantly less than when it was analog.

off topic for a moment or two:
Agreed, modern answering machines are mediocre. Pardon my ignorance, and the answer would vary with the country, but when did the telephone system (exchanges?) go digital?

> keep in mind that just like amplifiers we only have test kit that really measures the grossest of distortions.

I appreciate that THD is a flawed measure, (aside from a Geddes or other unobtainium measures) what’s the best?


Dave I’m really glad you’re in this thread.
While listening to pop/ rock music, 80% of the time I don’t pay much attention to voice – I listen to “the band”, esp the drums, so getting best midbass and transients is the priority that flows to me from that.

But for vocal based TV (HT), HT-type effects aside, clear voice is maybe 80% of what’s important. On reflection, 200~500 – 3500 Hz, I think that’s the sweet spot of many full rangers. Up to say $100/ pair, or maybe up to say $200 pair, what FR drivers best do that material and range?

Thanks
 

otto88

Member
2007-12-05 10:39 am
Magnetar

I’m not thinking of manufacturer’s specs, but Zaph Audio’s results. A limited selection, but imo a fairly good one, esp for moderate price drivers.

> There are low distortion drivers out but they are expensive.

I’m not fixed on low distortion, maybe CSD is a better measure.

But a driver that produces too much distortion for rock played at 105 dB peaks, may be fine for vocal based HT/ TV (drama/ comedy) averaging 85 dB, peaking at 90-95 dB. What level of distortion is “low”, is surely in part relative to usage.

> Bell Labs was right, use the lowest distortion drivers you can afford.

(Didn’t’ know they said that) but I’m looking for what’s appropriate for the application, a sweet spot. Not the best I can afford, but for spoken word, 80% good enough for a moderate outlay.

Cheers
 

otto88

Member
2007-12-05 10:39 am
Andy

> Any well designed HiFi loudspeaker will automatically have the necessary vocal acuity for HT.

I agree. But maybe more than enough acuity, ie beyond the point of diminishing returns. As above, we can mostly hear the other person's voice fine, via a single (50c?) phone speaker.

And simply going to a two or three way, with well done XOs ideally outside the key vocal range, should give big gains.

> Design for "good" NOT "adequate for the task"

I’m trying to design for a sweet spot of good listening and good value ;)

Cheers
 
otto88 said:
but when did the telephone system (exchanges?) go digital?

Here they are tossing the analog as fast as they can, all long distance is going over fibre, and all cel phones are digital (and they are quickly approaching half the phones)


I appreciate that THD is a flawed measure, (aside from a Geddes or other unobtainium measures) what’s the best?

Looking at the THD spectrum before it is collaped can be useful, but we are still in a world where the measures are a tool to help guide you, but you still have to trust your ears in the end.

Ans speakers measures... as Ed Schilling says, he can make a speaker measurement look almost like anything he wants. 2 measures will really only be compareable for instance with exactly the same physical set-up, the weather (ie temp, humidity, and barometric pressure) all need to be identical to have any hope, and the computer tools we use are limited. Further we tend to look at frequency domain data, but our ear/brain works in the time-domain *this is one of the reasons a FR with seemingly poor FR can still sound so good.

As Andy says it is pretty easy to get the basics pretty good. Him and i will disagree on XOs (i like to keep them far out of the critical 300-3k band, and will suggest that an augmented (or not) FR is going to be the king of voice,

dave
 

otto88

Member
2007-12-05 10:39 am
Hi Andy,

May be a misunderstanding, I’m not considering the *lowest denominator, but for value in a specific application which I don’t attribute as high a value to as I do to music.

I’ll first see if just L& R alone are ok, but suspect with a family of five (+ friends) and a wide room with seats up close, demands on the centre channel increasing, rather than my earlier thinking of just a "phantom" centre, we’ll be better with actual centre(s) . .

And if l do need a centre, I’m likely to want to ‘split’ the centre into mono L&R, either side of the screen, inboard of the left and right, so the voice drivers are all at the same mid-screen height, excursion is halved, etc

So I’m budgeting for probably four drivers, adding to my price sensitivity . .

Hi Dave,

If FR is the king of voice (augmented below say 250-300 Hz, maybe augmented with a super-tweeter) - you didn’t say which would be your recommendations at c $50, or at c $100 a driver?

At the lower price the FF85K, at the upper price the FE108EZ or enabled FF85K’s??

At present we’re using CRT, but the woofers Ive set aside for this aren’t shielded, so I need to locate an affordable LCD ~
That ‘frees up’ the driver choices, except for killing budget flexibility . .

If recommending your own drivers poses an ethical demur, feel free to contact me offline

Thanks
 
Remember, 300-3000 (or whatever similar range you pick) is just optimal for getting the information across, for intelligibility. Fidelity or tonal quality has little to do with it. I suspect back when they were actively researching the matter, their idea of low distortion was pretty high. Look at the distortion measuring equipment of that era, vs today. No contest. Understanding what a vocalist is singing, vs feeling that the timbre and texture is accurate and satisfying, is miles apart. IMO, even though male vocals tend towards the lower end of the spectrum, the clues that make it real are much higher up. For female vocals, the clues are higher yet, but without the lower registers, reality still suffers. I haven't built speakers in a while, but my experience was that driver quality was everything. No amount of crossover complexity, zobels, phase compensation, little magic dots, or anything else, will turn a bad driver into an excellent one. Maybe improve it, but not to a level I was ever happy with. Unfortunately, really superior drivers have become really expensive these days.
 
HT speakers IMO don't need the same attention as music speakers as they are usually set to small so don't do below 80Hz, mostly dialogue or steering effects etc.

When designing for HT the main things I consider are:
1. Tonal match of drivers
2. Kind impedance load for the HT amp as these are not flexible beasts for low ohms and their power numbers are over optimistic normally quoting for 2 channels driven and not for 6 channels
3. Run the mid woofer full range so all the main sound is produced by 1 driver which requires a driver with smooth roll offs and no nasties that require corrections
4. Fill in the top with a tweeter
5. Design for location
6. Power handling is lower considering the bass is all done by the LFE through a sub so a large driver is not required
7. If no off-axis viewing then a centre is not required
8. Centre to be a vertical TM

So how does this work in practice? See the pic below (from left centre, fronts, rears, centre rear).

1. All mid woofers are 850488 for the shielded version 850528 (now replaced by 830860 or 832873) but a Vifa P13WH-00-08 would have worked as well.
2. The speakers have an impedance plot from 6R5 to 8R5 which is amp friendly and can be driven by the weakest HT receiver.
3. I run the mid woofers full range and cross over to the tweeter 5000Hz+. This also has a benefit of no xo losses so the SPL is higher than with an xo... less power needed. These have an exceptionally smooth response and roll off so no correction is required.
4. As the tweeter crosses over after 5000Hz with a 2nd order xo, then a cheaper unit is all that is required and I used left overs... SEAS 27TFFC, Vifa D26NC-15-06, Vifa D26NC-55-06, Vifa TC20TD-05-06. All these tweeters have an Fs between 900Hz - 1500Hz (even the D26NC-55-06) so the high xo is beneficial.
5. Each speaker has been designed for the location. Fronts designed to go next to the plasma screen (do not have fronts too wide apart for HT as you need to lock to the screen). Centre fits under the screen and tilts up to the viewing height. Rears sit behind a sofa and fire sideways into walls with gives a null at the listening location and relies on reflected sound. The centre rear fires upwards to the ceiling and back wall relying on reflected sound.
6. No bass to do so a 5" driver is more than adequate. These little suckers can take heaps with an xmax of ±4mm and 110W capacity.
7. I allowed for off-axis viewing by adding a centre but can get away with phantom sitting on-axis.
8. I couldn't quite get a vertical TM due to the height restriction but it's very close.

So there you have my take on good clear sound for HT without breaking the bank.
 

Attachments

  • ht speakers700.jpg
    ht speakers700.jpg
    36.8 KB · Views: 457
otto88 said:
If FR is the king of voice (augmented below say 250-300 Hz, maybe augmented with a super-tweeter) - you didn’t say which would be your recommendations at c $50, or at c $100 a driver?

At the lower price the FF85K, at the upper price the FE108EZ or enabled FF85K’s??

I have been mightily impressed with the FF85K. Mind you we never bothered doing any searious listening before EnABLing them.

Althou it can be used as a "full-range" down to 100 Hz at limited volume (xMaxes at 100 with 1 W), this should be thot of as a 3/4" tweeter with a BIG surround that lets it be crossed as low as a couple hundred Hz. I really want to hear this in a shootout with some of the bug-buck tweeters. So if you can do an XO at 250-350 Hz it would be stellar.

dave
 

otto88

Member
2007-12-05 10:39 am
Conrad

> Unfortunately, really superior drivers have become really expensive these days.

The other day I heard it suggested that with cheaper drivers from China etc, and maybe more competition, that good audio is less expensive than ever. And we have innovation, eg XBL2, enabling, improving sonics

But I agree re singing voces, here I’m just looking at dialogue voices for intelligblitty


rabbitz & p10

I'll digest when time allows

Thanks
 

otto88

Member
2007-12-05 10:39 am
rabbitz,
That’s fantastic practical advice, thank you very much.

And the EnABLed FF85K sounds very promising. Dave, your thoughts on EnABLed FF85K v EnABLed CSS 125’s (both XOed c 250 Hz) for dialogue?

When I can get to it, a p10-eN FR v rabbitz MT recipe shootout would be great

What a forum!

PS
"the front row should be covered by the same drivers thru the critical range"
What is that range?