Audio Games

Audio Games

Part 1 - Microphones, Equalization, and Compression

I've had a long-running interest in "SSB audio" -- microphones, and the racks of "pro audio" gear associated with them. But I run a Yaesu FT-450 at 100 watts into a compromised antenna. I want to be understood more than anything else. So this article isn't about "extended" or "enhanced" SSB, with wide passbands, with the goal of "broadcast-quality" sound. It's about effective SSB -- clean, intelligible sound in a limited bandwidth.

Many of the problems in achieving that goal were covered in a classic eHam article -- "It's Not Your Microphone, It's You" -- by Steve Katz, WB2WIK/6. That's excellent reading for anyone picking up a microphone.

But it leaves open the question:

Does the choice of microphone make a difference?

There's a lot of mystification around that question. I realized that I had enough microphones in my collection to start to answer it, and share the findings. There's a description of the gear at the end of this article. I also wanted to see how the sound changed as it went through various stages of "shaping" that are common in the ham world. Rather than put probes into my rig, I used computer-based software to do typical audio processing jobs (there's no such thing as "too much control"), and recorded the results. All of the tests here are "wired". I'm still working on "over-the-air" testing -- you'll have to wait for a future article for some examples.

1. Microphones

There are many hams who say:

"I use the stock mic. The engineers at [choose the manufacturer] know what they're doing."

But those engineers had a budget to work with, and very little of it was assigned to the microphone. Some "stock" mics have such a bad reputation that there's a cottage industry devoted to fixing them. Bob Heil has a devoted following of hams who have found their stock mics inadequate. They can't all be wrong, can they?

It's very difficult to judge the sound of a mic with "real-time" monitoring. There's a lot of "bone conduction" in your head, and it's hard to distinguish the "mic component" from the "bone component" while you're talking. It's much better to record your voice, and listen to it played back.

I recorded a few microphones with minimal processing to see how much they differed. I suggest listening to the samples through a good pair of headphones. With most loudspeakers or cheap phones, you'll lose the low frequencies. Judge for yourself . . .

Behringer C-2

This (for me) is "gold standard" sound -- just a cardioid condenser mic with windscreen, 6" from my lips, off to one side, plugged into a digital recorder. No distortion, negligible coloration. I might not like it, and I might want to change it to go on the air, but it's me.

According to Shure, condenser mics have more trouble with breath noise than dynamic mics.

Sure enough -- when I close-talked the C-2, the factory-supplied windscreen was overwhelmed. Listening to the original WAV recording with Sennheiser HD-280 headphones, my eardrums could feel the push/pull of the high-amplitude, low-frequency breath sounds. They'd wreak havoc if they got into the audio chain. Use enough windscreening ! I haven't seen a ham station with a "StopIt" fabric windscreen, but we can always learn from the professionals.

Audio-Technica P615

This is a cardioid-pattern, dynamic "vocal mic". It's got two layers of foam inside the mic, and an added foam windscreen on the outside, so it's immune to breath noise and pops. I like it. It's the mic I normally use with the FT-450.

Yaesu MH-87 A8J

This is the stock FT-450 hand mic. I glued a piece of foam across the front -- without it, the mic is extremely sensitive to breath noise and pops. Yaesu doesn't specify what kind of element it has, or its directivity, but it's tiny, in spite of the "man-sized" housing.

Even with my added windscreen, this mic is bothered by breath sounds and plosives ("p", "t", etc). I've given it two chances -- one "head-on close mic'd", the second "half sideways", out of the breath stream. Both have problems.

Sennheiser PC-150

Sennheiser PC-150 mic, flat [click link to hear audio] This headset is designed for sound-card use, with a 1/8" stereo phone plug that expects +5 volts of bias on the ring. I was surprised -- there's plenty of low end, and the sound is nice. The mic position is just right, and (with added foam) there's no breath noise. A simple adapter lets it plug in to the FT-450. I used this mic for the equalization tests that follow.

You can do these tests yourself, with whatever you have around. The +5 volt bias on transceiver electret mics (e.g. many Icom's) matches the bias from most computer sound cards, so "rig mic to computer" adapters are simple to make. If you don't like the sound of your mic, get a different one. There's lots of choices between a stock rig mic and a truly professional studio condenser.

2. Bandwidth and Equalization

Once the microphone has done its work, the sound is going to be "shaped" -- its frequency spectrum, and its dynamic spectrum, will be modified. In a normal setup, that shaping is done by the rig's mic amplifier, equalizer (if present), "voice processor" (a compressor), ALC, and transmit IF filter. For inexpensive rigs, the manufacturers give very little information about the specs of those circuits. For expensive rigs, some information is usually available.

In an ESSB setup, or one with outboard audio processing, the rig's mic amplifier, equalizer, and compressor are replaced, and only the rig's ALC and IF filter are used. The characteristics of the outboard gear are known. Such gear isn't cheap, but it is now affordable (that is, it costs less than an entry-level transceiver).

"Voice Shaper" is a free Windows program that does equalization and compression. It's really easy to use, much simpler than many "pro audio" boxes. It's a good base for testing and demonstrating the effects of "post-processing".

"Voice Shaper" does have one problem -- latency. Digital processing (without DSP processing) takes time. On my laptop, there's noticeable delay between "speaking" and "hearing". That doesn't affect these recorded tests. I've used "Voice Shaper" to drive my FT-450 with good results, but I couldn't monitor myself.

Computer-based processing is only as good as the sound card on the computer. According to a recent QST article, "stock" sound cards are more than adequate for radio-quality audio work, so I was able to use my laptop computer for these tests. The soundcard sampling rate is about 11 kHz, so frequencies above 5.5 kHz will be lost.

2.1 Bandwidth Demo

This is what happens when lows and highs are filtered out. With almost 4 kHz of bandwidth, we're in "ESSB" country. My voice still sounds like my voice, in spite of the narrowed range.

This is "standard SSB" bandwidth -- 2.4 kHz wide. My voice isn't "my voice" any more, the tonal balance is considerably changed. But I'm still understandable, since there's enough of the high frequencies to reasonably distinguish between consonants. "F" is largely gone, but "sh" and "sss" are still OK. It's "telephone-quality" speech. I'll use this bandwidth for the equalization demos that follow.

2.2 Equalization

Equalization boosts some frequencies, and reduces others. Voice intelligibility (in English) depends a lot on high-frequency components of consonant sounds, less so on low- and medium- frequency vowel sounds. But the high-frequency components are weaker than the lows. So when the bands get crowded, and propagation is poor, it pays to "re-allocate" power to the high frequencies, and away from the low frequencies. We can do this with an equalizer.

"Voice Shaper" has an octave-band equalizer, with band-center frequencies of 300, 600, 1200, and 2400 Hz. I'm using the PC-150 as the mic. Let's play . . .

In this sample, the 300 Hz band is cut a bit, and the 2400 Hz band is boosted a bit. The goal is to sound something like a Heil HC-5 mic capsule. We've added "punch" or "presence" to the basic telephonic sound.

Here, the 300 Hz band is cut substantially, and the highs are boosted. We have left "natural sound" behind -- nobody sounds like this in real life. The is to sound something like a Heil HC-4 "DX" capsule -- not pretty, but effective.

This is downright ugly, but it will be comprehensible at low S/N ratios when all else fails. It's "pile-up busting" sound.

3. Gating and Compression

"Gating" completely eliminates soft sounds. With gating, when you stop talking, your rig stops generating power (even if it's in XMIT mode). It's handy to avoid transmitting fan noise from your amp while you're thinking. While you're speaking, though, the fan may be audible underneath your words.

The compressor takes soft sounds, and makes them louder. If 0 dB is full output, the compressor might take a sound at -10 dB, and boost it to -5 dB. We could call that "5 dB of compression", or "2:1 compression" (because an input change of 10 dB causes an output change of 5 dB). If a sound is at -20 dB, and the compressor boosts it to -10 dB, that would be "10 dB" or (still) "2:1".

"Voice Shaper" uses ratio measurement -- "2:1" -- and has a graphical display of input and output signals that makes the concept easy to understand and work with.

If the compressor is set up at a high ratio, it will take the sound of your rig's fan (or your boat's motor) and boost it to be almost as loud as your voice. So we should combine compression with gating to get best results. With "full-bandwidth" audio, the amount of compression is controlled by the high-amplitude, low-frequency components of the voice. That's not good. Some "pro audio" gear handles the problem by using "multi-band compression" -- one compressor for the highs, another for the lows. That's not an option for "Voice Shaper".

Compression is most useful in low S/N situations, when you want all the power, all the time. For those conditions, we'd want to equalize with our most severe "low cut, high boost" settings. Since the compressor is placed after the equalizer, the compressor will not be driven by the low-frequency components. They ain't there any more, killed by a -10 dB cut at 300 Hz.

Here is what audio sounds like with gating, but with no compression. It was quiet in the house when I recorded this -- no TV in the backround, and the rig was off -- so the effect is subtle.

This is a moderately compressed signal. There's more "punch", and the average energy in the signal is increased. The sound isn't natural, but it is tolerable.

This is a severely compressed signal. It's lost all the character of the natural voice, and nobody will compliment me on my "nice audio". But if I want to be heard -- and I don't care what I sound like -- it's the ticket to success.

This testing has changed my thinking about running my radio. I've been using the A-T P615, plugged into the rig, as my microphone. I'll have to wire an adapter and try some on-air testing, to see whether the Sennheiser PC-150 "sounds better" than the A-T P615. If it does, I will use the PC-150, with its mic plugged into the rig, as my preferred mic/headset combo, and "retire" that heavy, comforting A-T hand mic. I'll work out my choice of FT-450 equalization settings on the air. For my next contest, I'll be running Voice Shaper with heavy equalization and compression, to get the most out of my minuscule ERP. Whether you rely on the rig, or have outboard audio gear, the steps in shaping a voice are similar. I hope I've taken some of mystery out of it.

Equipment Notes (prices are approximate, in $US):

Microphones:

Audio-Technica P615
A cardioid dynamic "vocal mic", with a few dB of "presence hump" above 2 kHz. It's a less- expensive substitute for the Shure SM58, and I use it with an aftermarket foam windscreen (over the grille, which hides two internal windscreens). It's built to spec by A-T for the "Alliance of Independent Music Merchants", and it's available at independent music stores who are members of that trade association. $50 at Portland Music -- "This is the cheapest mic we sell". The Audio-Technica ATR30 has similar specs.

Behringer C-2
A phantom-powered, small diaphragm, cardioid "true condenser" microphone. It's compact, and will handle 140 dB -- it's happy recording 6" above a drumhead. Response rolls off gradually below 200 Hz (-10 dB at 50 Hz), and is flat above that to 5 kHz. $60 per pair. Behringer makes a standalone phantom power supply for condenser mics -- $20. For a rig with a built-in equalizer, one of these mic's and the phantom power supply might be enough for the "perfect sound".

Sennheiser PC-150
The PC-150 is a mid-range "VoIP and gaming" headset, with a cardioid (probably) electret (probably) mic. The mic is on a boom, about 1.5" away from the lips, slightly off to the left side. I have added a tube of foam (see below) as a windscreen (the factory-supplied windscreen is not thick enough). I'm surprised by how much I like the mic. The headphones are comfortable, with small earcups, and OK for radio use. They are not "hi-fi" phones, and sound isolation is poor. Mine was on sale for $40 (Cdn). Replaced by the Senn PC-1150.

Yaesu MH-67 A8J
The stock mic for the FT-450 transceiver. I can't find a circuit diagram, and Yaesu is mum on what kind of element it has, and what its pattern is.

Recorder:

Samson H4 -- an "all-digital" multitrack or stereo portable recorder, with phantom-powered preamps and built-in condenser mics. $300. All the recording and processing for this article was done in CD-quality format -- 16-bit WAV files, sampled at 44.1 kHz. For publication, they were compressed (once!) to 320 kb/sec MP3 files.

Software:

"Voice Shaper" -- a computer-based octave-band equalizer /noise gate / compressor by Alex Shovkoplyas, VE3NEA. This is really elegant software -- it does just what I need it to do, and it's got a great user interface. Available at www.dxatlas.com. Free.

Computer:

Dell Inspiron 1150, manufactured in 2004, with Windows XP SP 2. I used the stock SigmaTel 16-bit soundcard. There are some "digital artifacts" and granularity in my recordings. I think a faster CPU would eliminate them, and reduce Voice Shaper's latency time.

Windscreen Foam:

Foam paintbrushes are a good source of open-cell foam for windscreens. Use a razor knife to split the foam at its glue joint and remove it from the handle. Use silicone sealant for glue to fabricate the windscreen and attach it to the mic (if necessary) -- it doesn't dissolve the foam. See -- it's still ham radio.

Copyright Charles P. Cohen 2008

Member Comments: