My ears! My ears!

September 17, 2007 2:20 PM

Some sound design questions to follow up on this golden moldy oldie...

Red Book standard CDs seem to offer an output range of 20 to 20000 Hz.

• Can I publish a CD which can output mono frequencies between 20 kHz to 48 kHz through common CD players?
• Would I have to use SACD? In general, what would I use (hw/sw) to author an SACD?
• Are standard consumer speakers incapable of playing back > 20 kHz, or is it just that speakers aren't tested any higher for obvious reasons? (If I distributed a CD/SACD with said high frequencies, would consistent playback be impossible?)
• If no, would audiophiles have speakers capable of said playback range?
• If I used Max/MSP to generate pure tones, is my only frequency output limitation the audio adapter?
• If I recorded those (inaudible) pure tones to a sound file, what file formats could I use for storage?
• Are there caveats about the cycle~ MSP object with respect to generating tones > 20 kHz?
• If I wrote a Max/MSP Runtime program to generate tones, what sound adapter/speaker combination could I use for testing which would work with both Windows XP and Mac OS X, which would reliably output 20 - 48000 Hz?

Sorry if some or all of these are dumb questions. Thanks for your help!

posted by Blazecock Pileon to technology (14 comments total)

1. No.
2. I don't know - SACDs are rare, but you might be able to find the standards online.
3. It depends. Some claim they are. Few *really* are. *(see note)
4. Audiophiles would not only claim that they had speakers capable of reproducing it, they'd claim that they can hear it, and that it sounds better with a wooden knob. The reality is that they'd probably not get reproduced. *(note again)
5. My guess is that the first (of many) frequency limitations you'd hit would be in the D/A - I doubt that many D/As handle frequencies too high above 24K.
6. I want to say any file format, but I have reservations - if you're entirely in the digital domain, you'd have to find a format with the appropriate number of samples/second - they exist, as they're sometimes used in 96K protools sessions. My first guess would be .aiff - I know Protools loves aiffs.
7. I don't think anything is going to be reliable with respect to 48K - if you're really looking at recording stuff that high, you're looking at equipment that costs orders of magnitude more than you're willing to spend.



*And one more note about things like speakers - when I say that most speakers can't reproduce frequencies above X, that's not entirely true. It's more that they can't produce those frequencies within a reasonable tolerance from the rest of the frequencies - there's no brick wall at 20K that keeps speakers from making higher frequencies - it's just that as you get above 20K, those high end frequencies get much, much quieter.

Can I ask what you're doing this for? How high exactly are you planning on going?
posted by god hates math at 2:42 PM on September 17, 2007

6 (part 2). Now that I think about it, Adobe Audition claims to support some freakishly high sampling rates on record, so it's possible that it will attempt to save data at that rate as well. Were I still at work, I'd try to save a 96K .wav, but since I can't do that right this second, I'm going to go ahead and say that unless the .wav standard says anything about limited sample rates, you could use .wavs as well.
posted by god hates math at 2:52 PM on September 17, 2007

A standard CD has a built in limitation of 20khz, with a filter. SACD would work but I do not believe there are any cheap authoring options. DVD audio would seem to be the way to go, as would old fashioned reel to reel tape. As for speakers, so can, some cannot reproduce those frequencies. A supertweeter that will do the job is not that expensive. You have to make sure the amplification equipment also handles those frequencies, which is not uncommon.
posted by caddis at 2:59 PM on September 17, 2007

CD's are sampled at 44.1kHz, and CD players playing from audio CDs are designed exclusively for that sampling rate. However, many very modern sound cards are capable of 192kHz sampling rates.

At 44.1kHz, you can't practically go any higher than 20kHz, in fact you may not be able to get there.

The ear's limited ability of detect pure tones is not the only important psychoacoustic factor, when it comes to 'the absolute sound'. I've heard it suggested that the ear can detect the presence of frequency components well above 20kHz in non-steady state testing (not a pure tone, something dynamic like a cymbal being struck).
posted by Chuckles at 5:14 PM on September 17, 2007

Don't confuse sampling rate with frequency. Most standard sound cards can handle a 48kHz sampling rate but by the nyquist theorem that means the maximum frequency that can be reproduced is half of that, or 24kHz. And that's the theoretical maximum, achievable only by a perfect brickwall filter, so in reality it will be somewhat less. For example CD redbook has a sampling rate of 44.1 kHz which would lead you to believe that it could reproduce 22.05 kHz tones, but in reality it's limited to 20 kHz because we can't build a perfect aliasing filter.
posted by Rhomboid at 5:17 PM on September 17, 2007

As I said in the last thread: tell us what the damn project is already! It will be much easier to help you.

Can I publish a CD which can output mono frequencies between 20 kHz to 48 kHz through common CD players?

No. As mentioned above the sampling rate of CD's means they cannot represent these frequencies. More info can be had at wikipedia's synopsis of the Nyquist-Shannon sampling theorem.

The only animals that could possibly hear these tones anyway (at least towards your upper bound) would be those that have echolocation abilities.

If no, would audiophiles have speakers capable of said playback range?

Doubtful. 35khz is considered quite high for a tweeter, because it's debatable whether people can detect these consistently (You lose your higher ranges of hearing first, so if anyone can hear them it would be very young people).

If I recorded those (inaudible) pure tones to a sound file, what file formats could I use for storage?

Use a WAV file with a sampling rate well in excess of twice the highest frequency you wish to represent. Most sound cards have a maximum sample rate playback of 96 kilohertz, meaning your upper limit will be 48 kilohertz. If you need to go higher than that with pure tones, analog generation (simple oscillator circuit) will be easier.
posted by phrontist at 5:36 PM on September 17, 2007

Most sound cards have a maximum sample rate playback of 96 kilohertz, meaning your upper limit will be 48 kilohertz.

By which I mean computer sound cards.
posted by phrontist at 5:40 PM on September 17, 2007

The only animals that could possibly hear these tones anyway (at least towards your upper bound) would be those that have echolocation abilities.

The listening subjects are not human, and do not have or require "echolocation abilities" to hear these test tones.

Anyway, it sounds like CD is out, and either DVD-A, SACD or sample playback are possibilities.

Are special amplifiers required for "supertweeters"?
posted by Blazecock Pileon at 7:20 PM on September 17, 2007

An amplifier that can drive a hypersonic driver is easy enough to find/build, but you won't find it in Future Shop or Circuit City. It is actually much easier to produce high SPL as frequency goes up, because tweeters tend to be substantially more efficient than woofers. Caution though, because they also tend to be a lot more directional - people talk about certain tweeters being "like a laser beam".
Directionality will make the driver more efficient in the pointing direction, obviously, but I'm fairly certain that tweeters are more efficient for many other physical reasons as well (like practical cone size vs. frequency, and practicality of horn loading).

Burning DVD-A may be possible, barely, but I'm pretty sure a one off SACD is not at all practical.

You might be able to find an mp3 player that can handle a 192kHz sampling rate, though it will require a lot of luck. As I said before, sound cards and onboard audio solutions that can convert at 192kHz are becoming fairly common. You would still have to verify the output carefully though - you sometimes see reviews of sound cards which include the real world frequency response, or you could hook one up to a scope - I wouldn't trust a product just cause it says it handles 192kHz on the box :P
posted by Chuckles at 8:43 PM on September 17, 2007

Burning DVD-A may be possible, barely, but I'm pretty sure a one off SACD is not at all practical.

It sounds like the copy protection mechanisms for DVD-A and SACD would kill any chance of authoring a test-tone disc, the dearth of authoring software notwithstanding.

Perhaps a Max/MSP Runtime binary, coupled with the right audio adapter and amp/speaker combo, is a reasonable way to go.

Are there commercially available amp/speaker combinations?
posted by Blazecock Pileon at 9:05 PM on September 17, 2007

Plain old DVD-video supports PCM sound up to 96 kHz. That makes it entirely within the realm of possibility.

Most good speakers can probably reproduce tones above 20 kHz, but I imagine their response curve in that range is not very flat. You can always test this with the appropriate equipment (although I don't actually know what that equipment would be -- probably a sound pressure level meter of some sort).
posted by neckro23 at 1:01 AM on September 18, 2007

Harmon/Kardon have always specialized in high bandwidth amplifiers. Lots of other amps are also high bandwidth, they just brag about it less. Good, used Harmon/Karmon stuff can be picked up cheaply. ERT super tweeters can be added to your existing speakers.
posted by caddis at 5:21 AM on September 18, 2007

I was thinking more along the lines of Ultrasonic sensors, but since this problem appears to be fundamentally acoustic (heh.. is that clear, at all?), I think caddis is probably on the right track.

On the other hand, building something could be a lot cheaper. The LM3886 audio power op amp has a gain bandwidth of 3MHz, so presumably it could easily drive a transducer out to the 90+kHz of 192kHz sampling rate sound cards.
By cheaper I mean, about the same price if you have to contract someone to build it, but ~$50 (or less) if you know how to do it yourself.

Unfortunately, other than 'try a bunch of stuff', I don't have any cheaper alternative for the transducer..
posted by Chuckles at 7:04 PM on September 18, 2007

Thanks to all of you for your valuable input.
posted by Blazecock Pileon at 11:56 AM on September 20, 2007