How to fix DSP-100 problems (NEW) in Questions answered #3

I used to have my opinions about good audio and some of what other scientists and audio engineers have been saying about audio in human speech. Brian K3VR did a much better job than what I was saying here, so with his permission, I am re-publishing , what he published on QRZ.com. 

BANDWIDTH AND AUDIO TRENDS IN AMATEUR RADIO: AN HISTORICAL PERSPECTIVE

Let's review what we know about speech and intelligibility and the trend toward a better sounding paradigm in amateur radio. At some point in the early part of the last century, the telephone was produced to carry voice information. The telephone company research at that time indicated that carrying audio in the range of 300-3,000 Hz was a good compromise for telephone quality audio based on the equipment available at that time. Intelligibility in the range of something approaching 75%, as rated by listeners, was transmitted down the line using the twisted pair. That meant that 25% or more of the speech intelligibility was lost, but 75% was considered pretty good, considering the size and quality of the microphones and speakers used in the telephone's handset.

In the mid-50's, Art Collins demonstrated SSB to the military. Collins used telephone company research data to develop his sideband filters and he sold General Curtis LeMay on the idea that SSB should replace AM in the military. At that time, amateur radio operators followed the ARRL, who heavily promoted SSB operations in their magazine, called QST. The new SSB equipment was very expensive, and AM operators complained bitterly that the ARRL was trying to sell them a bill of goods by promoting a technology that sounded like Donald Duck.  Eventually hams realized that SSB was a promising technology that was more efficient over the long haul than AM, produced no heterodyne like AM, and took up only 1/3 the spectrum of typical AM operations.

Intelligibility research didn't suddenly end in the middle of the last century. Scientists, Medical Doctors, and Engineers in the medical, telecom, hearing, acoustics, internet, military, and music industries have all performed their own studies which have been replicated literally thousands of times, proving the validity of their research protocols. Psycho-Physio Acoustics is now a recognized medical sub-specialty along with Audiology. Many of these disciplines publish peer-reviewed research on a monthly basis in their own technical journals.

In this century, as well as the last, amateur operators have been fascinated by speech, intelligibility, and quality sound reproduction in amateur radio. The typical SSB bandwidth of 300-2,700 Hz was missing something for many of these operators, many of whom are engineers and experimenters. AM operators in particular have always been interested in preserving the intelligibility and fidelity associated with that mode before the advent of SSB. The trusty D-104 by Astatic wasn't good enough anymore. The Electro Voice 664 was very good, but it was pricey. A sound engineer for musicians and and amateur operator, Bob Heil, K9EID, invented a line of microphones for "rag-chew" and "dx" applications that were specifically designed for amateur radio. Many amateurs gravitated to the HC-4 and HC-5 elements in his dual use microphone line. Next, Bob invented the Gold Line, which soon became a very popular microphone. It was like a studio Mic., except that Bob purposely built in a peak in the response curve at +4dB at 2,000 Hz in order to facilitate its use on the amateur bands. The PR-20, PR-30 and PR-40 were released not long afterward. Amateur operators developed a seemingly insatiable desire to improve the sound of their transmitted audio.

Julius Jones invented the W2IHY line of equalizers. These devices are ham specific audio gear, which, unlike off-the-shelf, commercial gear, are shielded to prevent RF feedback. Average hams with no previous interest in audio became more and more interested in sounding as good as they possibly could on the ham bands, largely as a result of these products and the influence of the hi-fi AM movement, sometimes called the East-Coast sound. Some amateurs take audio experimentation to yet another level, employing whole racks of professional audio gear in their quest to sound better. Like dx enthusiasts who spend many thousands of dollars on very tall towers and monoband antennas to try to work the "rare ones," these amateurs are in pursuit of audio excellence. Some of them have achieved it, others are still working on it.

In the 1990's, many amateurs began to modify their SSB equipment in order to provide greater bandwidth in order to facilitate better sound quality. The Kenwood 950SDX, TS-850 with DSP-100, the Yaesu FT-1000D, and the Yaesu MkV are a few of the transceivers that have been so modified. Later, the Icom 756 Pro series of transceiver offered a 100~3,000 Hz bandwidth as standard equipment. Icom also included software based equalization (bass and treble control) to their line. Icom also included the ability to listen from 100~3,600 Hz in SSB and they provided amateurs with a receive bandwidth of 9 kHz in AM, instead of the traditional 6 kHz found in most other SSB oriented radios of the day. The trend continues. Off the shelf radios now offer even wider bandwidths -- some, like the Flex-Radio SDR-1000 offer bandwidths up to 10,000 Hz for SSB and up to 20,000 Hz for AM. Ten-Tec, long known for its 2.4 kHz transmit bandwidth filters now offers the Ten Tec Omni 7 with a possible transmit bandwidth of 80~4,000 Hz. Yaesu offers parametric equalization and balanced microphone (XLR) input connections in their high end transceivers. Many amateurs are still using the transceiver made famous by Art Collins (The KWM-2) which, with its 300-2400 Hz audio response, was state of the art in 1959. The popularity of the Icom 756 Pro line, Heil microphones, W2IHY audio accessories, and commercial audio gear like Behringer and Shure, in use today, attests to the popularity of audio experimentation and the quest toward "better audio."

HUMAN SPEECH

Modern research informs us that human speech is made up of vowels, consonants, and harmonics which can range from 80Hz at the low end of the bass spectrum, all the way to 8,000 Hz at the higher end of the frequency spectrum. Speech is made up of fundamental sounds and their harmonics. Speech sounds are classified into phonemes, which are grouped into the following basic categories:

Vowels: Monophthongs and Diphthongs; and Consonants:  known as: Approximates, Nasals, Fricatives, Plosives, and Affricates.

The plosives, p,b,t,d,c,k,q and g are produced in various regions of the mouth and throat, e.g.; bilabial, alveolar, palatal, velar, and uvular. The nasal sounds m and n are produced in the bilabial and alveolar regions. The trills, B and R are produced in the bilabial and uvular sections.

The fricatives, f, v, x, and H are produced in the labiodental, velar, and glottal regions, and the approximate, j, is produced in the palatal region, while m is produced in the alveolar region.

Every phoneme has its own pattern in a spectrogram. Phoneme spectrogram signatures for various speech sounds include lots of concentrated energy called formants. The formants present as groupings of different frequencies depending on where the formants are formed in the mouth and throat. Human speech is a series of buzzes (voicing) pops (fricatives) and hisses (fricatives). All of these mouth and throat sounds are used together to create intelligible speech.

THE IMPORTANCE OF BANDWIDTH

Fully 25% or more of speech intelligibility is lost if we fail to include frequencies above 3,000 Hz. Speech has been successfully rendered below 3,000 Hz for years, but such speech is considered to be pinched and unnatural sounding by many people who enjoy hearing the entire, natural content of human speech. Some amateur operators have attempted to replicate natural bass and treble sounds within a 3,000 Hz bandwidth. This has sometimes produced an unnatural boost in bass and treble notes in an attempt to artificially restore that which has been taken away with reduced bandwidth. This process sounds no more natural than the audio it attempts to replace. It's different, and possibly richer tonally, but it doesn't compare to listening to the human voice with its full range of intelligibility, between 80 and 8,000 Hz.

A pair of amateur radio operators who feared that audio and bandwidth experimentation would limit bandwidth available for narrow bandwidth ssb (NSSB) users petitioned the FCC for rule changes. In 2004, the Federal Communications Commission DENIED a petition that sought to limit the bandwidth of SSB and AM transmissions. The FCC called the petition inconsistent with the goals of experimentation and flexibility in the amateur service. The FCC also said the petitioners had failed to convince them that any changes were needed. Two years later, in October, 2006, the FCC issued an Omnibus Rule and Order that expanded phone bands on 40 and 80 meters, greatly relieving the congestion about which users had long complained.


WHAT'S MISSING IN BANDWIDTHS BELOW 3,000 Hz?

We are missing approximately 25% (or more) of the intelligibility contained in normal human speech. The intelligibility factor is important when hearing loss is present (as is the case with most middle-aged adults). As we age, we experience presbycusis, the inability to hear higher pitched tones. High school students have discovered that they can use a high frequency sound (17 kHz) as the ringtone in their cellphones. Older teachers are unable to hear the tone and the students use the ringtones in class to alert them to the fact that they have new text messages. The "mosquito" ringtone enables students to engage in a form of hi-tech note passing.

While it's possible for young people to hear tones as high as 17,000 Hz, it's generally accepted that the useful elements of human speech are contained between 80 and 8,000 Hz. When using low bandwidths, such as those inherent in telephone transmissions and NSSB, consonants with high frequency components (/f/ /th/ /s/) may sound very annoying. Some consonants (/d/ /g/ /k/) and consonant combinations (/dr/ /gl/ /gr/ /pr/ /spl/) are highly intelligible with natural speech, but problematic in narrow bandwidths. Other difficult combinations in narrow bandwidths are, for example /lb/, /rp/, /rt/, /rch/, and /rm/. The Harvard Psychoacoustic Sentences test is a closed set of 100 sentences developed to test the word intelligibility in sentence context. The first five sentences of the test material are:

The birch canoe slid on the smooth planks.

Glue the sheet to the dark blue background.

It's easy to tell the depth of a well.

These days a chicken leg is a rare dish.

Rice is often served in round bowls.

Results using large sample groups of human subjects using many tests similar to the one above indicate that bandwidths of 8,000 Hz provide almost 100% accuracy of interpretation vs. only 75% accuracy with bandwidths to 3,000 Hz.

IMPLICATIONS FOR AMATEUR RADIO, MEDICINE, AND TELECOMMUNICATIONS IN GENERAL

Wider bandwidths are legal and increasingly popular on the amateur radio bands. The FCC has also stated that experimentation balanced with consideration for others is encouraged. In the medical field, experimentation with wider bandwidths has had direct benefits for those with hearing loss. Modern hearing aids are designed not simply to amplify, but to enhance frequencies that have been lost or reduced in the hearing of the user. In the field of streaming audio for use over the internet, the G.722.1 protocol describes a digital wideband coder algorithm that provides an audio bandwidth of 50 Hz to 7 kHz, operating at a bit rate of 24 Kbit/s or 32 Kbit/s. In the governmental arena, the proposed standard for the FAA and Military communications is a data/voice protocol called VDL-3 which is being jointly produced by the telecom giant Harris, Rockwell Collins, and others. The major benefits of this protocol are said to include "AM-like voice quality only better." The voice channels for this protocol are 6.25 kHz wide. A similar protocol used in air traffic control in Europe is 8.3 kHz wide. Both systems recognize the need for more intelligible, quality based audio by replacing an outdated 3,000 Hz SSB-type bandwidth standard with something more than twice as wide.

THE FUTURE?

The benefits of using bandwidths more compatible with human speech are undeniable. Reduced listener fatigue, increased fidelity, greater intelligibility, and more natural sound are just a few of these benefits. With modern research, pinched and robotic sounding "space shuttle" audio is becoming a relic of the past. Modes like Digital Radio Mondiale's (DRM) and AOR's ARD-9800 digital voice protocol may represent the building blocks of a new direction for the future of amateur radio. Digital voice communications users such as ARD-9800 report no noise level, speech with an FM broadcast-like quality, and the ability to communicate using less power. As more useful digital voice modes evolve, they are likely to be adopted by those searching for lower noise and increased fidelity. 

 Hints, Tips, and Recommendations for Amateurs Experimenting With Increased Fidelity/Intelligibility SSB

1. A receiver section employing a crystal filter system like a Collins KWM-2 will not hear any bandwidth change beyond 2,700 Hz, or below 300 Hz, by virtue of the "brick wall" filtering system. Many radios in use on the amateur bands today will be "deaf" to improvements in audio fidelity and intelligibility due to receiver design parameters.

2. In contrast to #1, a receiver like that found in an Icom 756 Pro III can detect and appreciate bandwidth changes between 100 Hz and 3,600 Hz. Therefore, improvements in intelligibility and fidelity can be detected out to the maximum receive bandwidth of 3,600 Hz in SSB. Likewise, a Kenwood TS-850 with DSP-100 can detect bandwidth changes resulting in improved intelligibility out to approximately 6 kHz. Many other radios currently in production can also detect increased bandwidths.

3. An improperly adjusted carrier point in a radio like a Yaesu MkV will cause an increase of transmitted energy on the unwanted sideband, adding nothing to fidelity or intelligibility, but vastly increasing interference potential to other spectrum users.

4. Too much audio preemphasis may cause IM3 and IM5 products to increase, causing potential for interference. It is the transmitting station's responsibility to monitor IMD carefully.

5. Receiving stations with noise blankers engaged will often complain of "splatter," when in fact, the noise blanker is adding unwanted artifacts to the recovered audio in the receiver.

6. Stations operating receivers in close proximity to other stations will often complain of "splatter" when in fact they are simply attempting to operate within the transmitted bandwidth envelope of the other station's transmitter. Let them know your transmitted bandwidth is wider than 3 kHz. Determine whether or not a frequency change is in your best interest.

7. The goal of achieving fidelity is sometimes at odds with the goal of achieving intelligibility. Because the energy present in the human voice drops off as frequency increases, preemphasis may be required to bring the level of certain consonants to a point where other stations (with appropriate equipment) can hear them with the desired intelligibility. Preemphasis decreases fidelity but it may improve intelligibility, particularly in high noise environments.

8. High Fidelity audio has traditionally represented a flat audio response, between zero and 20,000 Hz. Audio transmitted in narrower bandwidths may benefit from the use of preemphasis with the goal of improving intelligibility when wider bandwidths are impractical. The benefits of higher fidelity transmissions are only practical when, A.) Space is available for such transmissions without causing interference, and, B.) When atmospheric noise levels are low enough for the receiving station(s) to appreciate increases in fidelity.

9. Amateur stations employing preemphasis should use sufficient detection methods (oscilloscopes, spectrum analyzers) to be certain their transmitters are not the cause of harmful interference and spurious emissions. [See note] Any kind of audio processing, compression, or equalization scheme, including the predetermined +4dB rise in many microphone elements could conceivably contribute to harmful interference. Ensuring the transmission of a 'clean signal' should be considered a primary part of "good amateur practice."

10. Any introduction of nonlinearity (overload to the point of distortion) to the transmission system, whether in audio or RF amplification, will potentially result in greater interference potential. Grid driven tubes or improperly filtered 12v transistorized amplifiers, for example, are inherently less linear than amplifiers driven by triodes. The RF output of amplifiers should also be monitored.

11. Improper tuning of linear amplifiers, overuse of preemphasis, introduction of nonlinear audio, overuse of compression and/or processing, "turning up" the power output of amateur transceivers (creating nonlinearity) in order to provide higher drive for RF amplifiers, and RF feedback (antenna problems) will also present greater interference potential.

12. Increased low frequency response means that stations who are even 10 Hz "off frequency" from the receiving station will sound unnatural. Stations experimenting with transmitting low frequencies (under 100 Hz) are encouraged to employ transmitters using strict frequency tolerances. A TCXO referenced to WWV is highly recommended.

The Bottom Line: Major increases in SSB intelligibility and fidelity are possible. The responsibility for interference mitigation and running a clean station rests solidly with the station doing the transmitting. Amateur stations experimenting with higher fidelity/intelligibility must ensure that their transmissions comply with accepted standards of spectral purity.


[Note] Spurious emissions include harmonic emissions, parasitic emissions, intermodulation products and frequency conversion products. Make sure your station's transmitted signal complies with FCC rules!


 Now as Paul Harvey would say the REST OF THE STORY.  W3OZ

More and more I am hearing about, and getting emails from guys buying a great new rig, that has an internal EQ and pretty good abilities to tailor ones own audio without external audio devices. Here is a typical scenario. A ham gets a nice new Icom Pro III. He has the ability to have a stock bandwidth on transmit of from about 100Hz to about 3 KHz. It also has the ability to have a more medium band pass and a narrow band pass. If the ham only does DXing or Contesting the medium and narrow EQs should work fine set up for 2.4 KHz or less. But he also wants to have the rig set up to sound great chatting to his buddies as well, he will have the EQ on the wide mode most of the time. This is pretty typical. He will not be busting pileups or anything like that just chatting on a regular frequency as he has done for 20 years or more.

Instead of buying a new microphone that will match the capabilities of his new rig he keeps the same old microphone he has been using for the past 20 years. Now he gets on the air and tries the new rig out on his buddies. He sounds pretty much like he always has but his buddies tell him what fine audio he has on this new rig. We all want to hear things like that after spending 3K on a new rig. But secretly he wonders why when he puts his rig in the wide mode he does not hear that much change and he fiddles with the EQ settings and wonders why he can’t get it to sound much different.  The reason of course is that he must remember that you will only have audio as wide as the narrowest device you have connected to the rig. It does not matter that you have your rig set up for a band pass of 100 Hz to 3 KHz it will be limited by the narrowest device. In many cases that offending device is the microphone.

Very popular microphones in use today are the Heil line of dynamic microphones. They are wonderful for working DX and in headsets for contesting. But here is the problem. The HC-5 cartridge that is in many of them, rolls audio off that is below 300Hz. and what is called the DX dream microphone, the HC-4 rolls everything off below 600 Hz.  Remember what I told you earlier about the fundamental voice frequency of an adult male voice being between 85 and 155 Hz? You can see that if you use these microphones in everyday conversation on the bands, you, as apposed to the audiophiles, are modifying your voice and are not sounding natural. This is normally a common complaint of non audio hams about those of us that are trying to sound better and are using enhanced audio (EA).

It is much better to get one of Bob Heil’s newer microphones like the PR-30 or 40 and then narrow the audio down, if you like, by using the built in resources of the brand new rig you just bought. You will lose nothing in the DX or contesting world but you will have the ability to sound better in your normal rag chew activities. Of course your friends are going to have to get used to your new sound and will not like it for a time, but as they get smarter they will turn around.

You see, in the past we have not had these new DSP type rigs with all the features we have today. The only thing we could do was to try to boost at about 2 K with microphones like the HC-5 and 4 to make the articulation better. We had no other choice then. Over the years we have gotten used to hearing our friends on SSB sound like girls in a can. We used to lament that the SSB audio we had was no way near as good as what we had in the AM days but slowly we got used to the crimped up tin can sounds that we have most of the time today. But the good news is that now, we can sound almost as good as AM with far less band space. No matter if you want to go the ESSB route or the more moderate EA enhanced audio route, it should be your goal as a good amateur to do your best to sound your best on the air at all times.

Back to the home page.