 |
 |
|
|
|
|
|
|
||||||||||||||||||
|
|
|||||||||||||||||||
|
|
|||||||||||||||||||
|
|||||
|
|
Document 28
The Technology Task Force continues oversight of the Digital Voice Working Group, High Speed Digital & Multimedia Working Group, and the Software Define Radio Working Group.
The Digital Voice Working Group report is included at Attachment A.
Startup progress for the SDR & HSMM Working Groups has been delayed due to a number of events, most prominently 9/11. Working Group members are:
| SDR | HSMM |
|
Bob Larkin, W7PUA (Chairman) |
John Champa, K8OCL |
|
Gerald Youngblood, AC5OG |
Dominique Delerablee, F1FRV |
|
Leif Åsbrink, SM5BSZ |
Dennis Silage, K3DS |
|
Gary Barbour, AC4DL |
Jim Idelson, K1IR |
|
Douglas Smith, KF6DX |
Neal Zipper, KR4IZ |
|
Paul Rinaldo, W4RI |
Paul Rinaldo, W4RI |
At Minute 47 of the 2001 Second Meeting the TTF was directed to develop standard guidelines for data collection, storage, and analysis to be used for 160 meter beacon data. The committee has no recommendation at this time regarding this item.
Respectfully submitted,
Joel Harrison, W5ZN
Chairman
ATTACHMENT A
Third Report of the ARRL
Digital Voice Working Group
to the ARRL Technology Task Force
January 2, 2002
1. Introduction.
The Digital Voice Working Group held a teleconference on December 15, 2001 to discuss:
A. The two QST articles on digital voice,
B. Information about those interested in digital voice in other areas of the world,
C. APCO Project 25, digital audio broadcasting (DAB) and other systems,
D. Voice quality evaluation standards,
E. The status of a digital voice kit,
F. Future work, and
G. Recommendations to the Technology Task Force (TTF).
2. QST and QEX.
We achieved our goal of obtaining exposure for our digital voice effort through two articles accepted for publication in QST. They appear in the January and February 2002 issues. The first article focuses on the history and contemporary theory behind digital voice modes; the second, on what is the state of Amateur Radio art. Those articles should generate significant interest and feedback. In addition, we have made a renewed call in QEX for articles on digital voice.
3. Digital Voice Around the World
Group member Charles Brain, G4GUO, continues his work on transmission protocols and error-correction schemes. That work may benefit regular data modes as well as digital phone. His work also involves software-based voice coders that use very low bit rates. He states he would like to investigate spatial diversity reception with digital voice. Charles has a standing invitation to write about his further studies for QEX.
Matt Ettus, N2MJI, has already done some work on the spatial diversity question, both on transmit and receive. He is using an orthogonal frequency division multiplexing (OFDM) modulation scheme and is experimenting with MELP2400, the replacement for government-standard LPC-10 voice coders. Matt has a short piece on his work in the 20th DCC Proceedings. He has been contacted about expanding that article for QEX.
Contact has been established with Michael Schulhof, K1OKI, who is pushing a system for HF digital audio broadcasting (DAB) designed by Thales, a French company (formerly Thomson-CSF). Michael was recently Chairman and CEO of Sony America and he now runs a private investment firm in New York. He states that Thales' participation in the making of an ITU standard for DAB has been intense, but it is not clear whether their system will see widespread use by international broadcasters. ITU standards currently describe systems from Digital Radio Mondiale (DRM, Italy) and iBiquity (USA).
The Thales system has been picked up by certain French hams who like it for Amateur Radio experimentation. Michael can arrange for Thales equipment to be shipped to ARRL for testing and transatlantic trials. According to him, the Thales gear has been tested in an occupied bandwidth of 3 kHz, which makes it a candidate for Amateur Radio. KF6DX is pursuing arrangements for those tests. The system is complex in its preferred embodiment and it remains to be seen whether amateurs will latch onto it.
Some news is starting to come from Japan. Earlier this year, both Kenwood and ICOM showed prototype 23-cm digital voice gear at a Japanese hamfest. An article appeared in CQ Ham Radio (in Japanese) about the show, a translation of which is posted on our digital voice page (www.arrl.org/tis/info/digivoice.html). The ICOM system uses code-book-excited linear-prediction coding (CELP, ITU standard G723.1) and a Gaussian minimum-shift-keying modem (GMSK). The Kenwood system uses advanced multi-band excitation coding (AMBE, from Digital Voice Systems, Inc., www.dvsinc.com) and a GMSK modem. 23-cm data-terminal equipment was also shown.
The above-mentioned equipment allegedly conforms to a JAIA standard called "D-Star." Since the ICOM and Kenwood units use different voice coding schemes, it appears they are not compatible with one another. Further investigation is needed to resolve that apparent contradiction.
We added Mr. Yoshi Nishimura, JA6UHL, to our e-mail list. Yoshi expresses a genuine interest in digital voice development. He indicates he is in good touch with other radio manufacturers in Japan and also with Japanese ham-radio publishers. He says no equivalent to the Digital Voice Working Group exists in Japan to his knowledge.
4. APCO 25 and Alinco
In November 2001, our group received information from League executives that Harold Reasoner, K5SXK, and friends were using APCO 25 radios through an Amateur Radio repeater in the Fort Worth, Texas area. APCO 25 is a national standard for the public-service community that provides digital voice and messaging services. In December 2001, Mr. Reasoner loaned ARRL a pair of Motorola APCO 25 units, one mobile and one hand-held, fulfilling another of our goals. Group members conducted an initial evaluation and the units were sent to the ARRL lab for further investigation. The units provide a good opportunity for testing the voice quality evaluation standards proposed below. Mr. Reasoner paid a personal visit to group members Barbour and Smith recently. He stands ready and able to further assist us.
Alinco is not the first with a digital voice system over ham radio, but they are the first to offer a hand-held and mobile digital voice system commercially in the Amateur Radio market, as far as we know. As of this writing, the ARRL lab has DJ-596 Alinco hand-helds that can be fitted with digital voice modules, but the modules themselves that were ordered have not yet arrived. The Alinco system uses continuously variable slope-delta modulation (CVSD), which has been employed in several prior Amateur Radio digital voice systems (see February 2002 QST). CVSD is inexpensive and relatively immune to errors in the data stream, but it generally does not reproduce voice as well as AMBE at identical bit rates.
5. Voice-Quality Evaluation Standards
Our group has reached a consensus that among the most important things it can do is establish a uniform standard for evaluation of voice quality. Any such standard should provide a basis for comparison among various analog and digital voice systems. A lot of work continues in this area commercially by the cell-phone and recording industries and we are able to draw on that work extensively.
Our recommendations are based on the concept that what someone hears or doesn't hear can only be determined by asking questions of the listener. ITU are working with KPN Research (Netherlands) and British Telecom to develop a quantitative standard for machine measurement of voice reproduction quality; but objective measurements must ultimately relate to the opinions of listeners. The widely accepted rating system is called mean opinion score (MOS).
The MOS system asks listeners to rate voice quality on a scale 0-5. Non-integer scores like 3.8 are possible. A score of 5 means that voice quality is unimpaired; a score of 0 means the voice signal is not understandable and is unusable. The system takes into account all means of degradation and does not place any constraints on how the listeners judge impairment. Audio engineers have used it for at least 10 years and a large body of MOS-rated audio samples exists in the public domain. Some of those samples can be found by following links on our digital voice Web page.
To remove any bias during investigation, an audio sample for evaluation is normally presented to the listener in one of two intervals, at random. The other sample can be one of known MOS and a comparison solicited from the listener. It is kind of like going to the optometrist: "Better, worse, same?" Many listeners participate and the results are averaged, arriving at a mean. The effect of individual listening talents is thereby minimized using MOS.
It is clear that objective measurements, like those being worked on by ITU, win over subjective by a long way when many rapid measurements must be made in a short interval. It is quicker and easier to record data than to ask many listeners what they hear. We expect that as far as human-hearing traits can be identified, they can also be formulated. Those benefits of objective voice-quality evaluation methods remain to be realized, though, at the time of this writing.
When bit error rate (BER) can be measured on a target system, sensitivity of the system is best specified with respect to MOS vs BER. When BER cannot be measured, sensitivity can be specified as MOS vs signal strength. We feel it is neither useful to use traditional ARRL methods to determine minimum discernable signal nor to use quantitative measurements involving tones. Other typical measurements, such as image rejection and interference immunity may be performed using the regular methods.
6. Recommendations.
The Digital Voice Working Group makes the following recommendations:
A. Adopt a uniform voice-quality evaluation system based on MOS and direct that it be used in all product reviews of digital voice systems. The simple criteria for MOS rating are as follows. Non-integer scores are possible.
| MOS | Quality | Impairment |
|
5 |
Excellent |
Imperceptible |
|
4 |
Good |
Perceptible, but not annoying |
|
3 |
Fair |
Slightly annoying |
|
2 |
Poor |
Annoying |
|
1 |
Bad |
Very annoying |
|
0 |
Unusable |
Total |
A. Use BER and signal strength vs MOS to find minimum discernable signal levels.
B. Direct the ARRL lab to evaluate Alinco gear and the APCO 25 radios being sent to them. Require a written report of the results by June 2002. Direct W1AW and League staff to support testing and transatlantic trials of Thales digital voice equipment. KF6DX will generate a test plan and coordinate.
C. Inquire of the IARU about the uniformity of rules and regulations pertaining to digital voice in various countries. Also ask about DAB and what impact it may have on Amateur Radio.
D. Consider the use of IEEE 802.11b digital networking equipment to occupy the 33-cm, 12-cm and 5-cm bands using voice over Internet Protocol (VoIP) and data modes before commercial interests overrun those bands.
E. Ask the ARRL President to appoint members to high-speed digital networking and "software-defined" radio working groups to begin coordination with this group.
7. Future Work.
We shall continue to investigate robust digital voice modes for Amateur Radio. It is our belief that digital voice modes must exhibit superior performance over their analog counterparts to become popular. Ancillary data and messaging services, as outlined in our last report, will tend to make digital voice systems more attractive to users. Digital voice equipment that can be retrofitted to existing transceivers still seems a good way to get more hams involved.
We recognize that standards for digital voice are emerging and our primary goal is to encourage and coordinate their use. We will do that by opening the information gates and continuing to reach out to interested parties around the world. We will inquire about the Japanese D-Star effort and others. We believe there is room for more than one digital voice method in Amateur Radio and that popular use and market factors like cost will ultimately determine what is successful. Interest is building rapidly but there will always be those who do not appreciate digital voice.
We renew our goal of developing a digital voice kit that includes a modem. We obtained new pledges from members and from at least one outside party to work on a software-defined modem for use with a digital voice kit. A software modem is attractive because potentially, it does not involve additional hardware. Some code has been written and we are near the debugging phase. At least one member has volunteered to put together a GMSK modem in hardware for VHF-and-above use. That modem would be offered as a stand-alone kit.
We are finding that this working group is generally ahead of similar efforts elsewhere. We shall maintain and add to our Web page www.arrl.org/tis/info/digivoice.html as events dictate. We want it and the committee reports page to be the places to which curious hams are steered for more information.
Respectfully submitted,
ARRL Digital Voice Working Group:
Gary Barbour, AC4DL
George Bednekoff, AC5WO
Charles Brain, G4GUO
John Gibbs, KC7YXD
Jesse Morris, KC5GTK
Doug Smith, KF6DX, Chair
Mike Tracy, KC1SX, HQ Liaison
1-2-2002
