The more I look at the world of radio in the traditional sense, the more it feels under threat. As a relatively late-comer to the shortwave and HF radio listening hobby and amateur radio, I’ve had the displeasure of seeing the closure of a number of services, relatively tough propagation conditions, increasing noise-floors in urban environments and increasingly antenna-hostile communities. More than that, I also see a dwindling number of hobbyists interested in the field, with some of the elderly making an involuntary exit as their health overtakes their interest.
In this environment, I felt it was a good idea to make some effort to preserve some of what makes shortwave and HF interesting. This included articles about DRM broadcasting, WebSDR/KiwiSDR receivers and collections of signals such as WeatherFax, VOLMET and NAVTEX. When I looked down the list of services I haven’t looked at preserving, the one that stood out to me was HF Weather Voice services.
HF Weather Voice
For mariners at sea, HF weather voice services provide critical weather information without the requirement or expense of satellite equipment. This often entails long broadcasts with weather forecasts being read out over the air periodically, by human or by machine and sometimes in more than one language, along with the characteristic noise and fading that often accompanies HF SSB transmissions. This service is an alternative and often complements the weather fax which provides graphical information and NAVTEX which provides textual information. The main advantage to weather fax and NAVTEX is that unattended reception is possible and routinely used whereas voice is often used in attended operation only (although recordings are possible).
Luckily for those who are interested in HF Weather Voice services, there is a long (although possibly unmaintained) list of service frequencies and times listed on this HFUnderground page. Unlike shortwave broadcasts which often change frequencies on a seasonal basis, most utility services maintain their frequencies consistently for many years to avoid inconvenience. As a result, while the listing could be somewhat dated with some of the transmitting stations no longer on the air or no longer transmitting at a given time-slot, it should still be reasonably accurate with regards to frequencies.
The Preservation Effort
I started my efforts with the data from the HFUnderground Marine Weather Transmissions page. Extracting the list of transmissions, I sorted them by country, aiming to receive a transmission from as many as I could. From the list of available frequencies for each country, I selected a frequency (or two) which would be most likely to be received (i.e. more transmissions, in a band good for propagation around the clock).
Reception was performed using KiwiSDR receivers which are publicly available on the sdr.hu website, using the kiwirecorder.py program. Audio settings were set for uncompressed audio, USB demodulation with a 20Hz to 3000Hz passband. Recordings were taken at the KiwiSDR receiver most proximal to the expected transmission site, although alternative receivers were used in case of strict time limits or poor antenna performance. Recordings of the most promising frequency were made for a period of between 12 to 24 hours on my VPS server (as my home internet is unstable), segmented by hour and losslessly compressed for transfer to my desktop.
Once the compressed files are collected on my desktop, I used an audio editor with spectrogram view to identify periods of voice transmissions (if any). Recordings of the best voice transmissions (in terms of signal-to-noise ratio) from each country were extracted from the recordings, resampled to 8000Hz and then exported as an AAC-HE audio file. FFmpeg was also used to generate an accompanying spectrogram image.
Should nothing be heard, another attempt would be made for a given country by choosing the next most promising frequency. Otherwise, a second receiver would be chosen to attempt reception. Multiple KiwiSDR receivers would be engaged in parallel within the October to November period for this first attempt, although I would not open more than one stream to any given receiver to avoid denying service to other users. However, in some cases, it was difficult to receive anything due to the low frequencies used by some stations, a lack of nearby receivers, limited transmission coverage area, presence of interfering signals, unreliable connectivity on the receivers or possibly due to service discontinuation or difficulties. Because of the extensive number of frequencies and countries, monitoring efforts took quite a lot of time over a period of two months, with processing taking some time over another month.
Having never heard the HF weather voice services of stations other than just my local neighbours (i.e. Australia, New Zealand, Thailand and USA), this project gave me the opportunity to listen further abroad and experience just how the other services sound like. However, because of the methodology and positioning of the available receivers, the quality of the reception may not be ideal in all cases.
The Recordings
The following section contains the gallery of successful reception recordings sorted into sub-headings for each country, sorted alphabetically. The information within the recordings is only accurate for the date and time it was received and is not kept up to date and should not be used for navigational purposes. Copyright in the transmitted information is believed to belong to the transmitting agency, however, the recordings are made by myself utilising the publicly available KiwiSDR receivers. They are provided here for documentary purposes, to illustrate the nature of HF marine weather voice transmissions from various countries at the listed dates and times.
Click on the spectrogram image or the link underneath to hear the audio.
Australia
Transmissions from Australia come from two transmitters – VMC (East) and VMW (West) with the information conveyed by a recognisable synthesized male voice.
This recording was received 20191101T2023Z_8176kHz and is an informational broadcast informing listeners of the weather voice frequencies in use.
This recording was received 20191101T2030Z_8176kHz and provides the forecasts, warnings and coastal observations.
Belgium
Transmissions from Belgium were received from Ostend Radio, transmitted by a real person. Transmissions are preceded by a tone and the securite call to all ships.
This recording was received 20191108T1426Z_2761kHz with new maritime safety information 673.
This recording was received 20191108T1433Z_2761kHz with repetition of maritime safety information.
This recording was received 20191108T1435Z_2761kHz with new maritime safety information 673, read by another operator.
Canada
Transmissions from Canada appear to use voice synthesis and are frequently made in both English and French. Transmissions are prefixed with a securite call to all stations, with seemingly different voices for each transmission station. The only exception appears to be navigation warnings, one of which appears to be voiced by a live human.
Transmission recorded 20191120T0844Z_2749kHz, consisting of transmission in French followed by English. While the transmitting station is identified, I could not determine what was actually said.
Transmission recorded 20191120T0937Z_2749kHz, following a similar bilingual format.
This transmission was received from Halifax Coastguard Radio on 20191119T2119Z_2749kHz in English, followed by French, including weather synopsis.
A transmission was received from Sydney Coastguard Radio on 20191119T1319Z_2749kHz with a navigational warning announced by a live person.
Otherwise, Sydney Coastguard Radio is synthesized in much the same way – this very long recording at 20191120T0737Z_2749kHz with transmissions in English and French.
Canada also has transmissions from Trenton (?) for the Pacific region using a different-sounding system. This recording from 20191113T1250Z_2054kHz.
Chile
Transmissions from Chile appear to be done by a live person in Spanish only.
Recording at 20191103T2101Z_4357kHz.
Recording at 20191104T0007Z_4357kHz.
France
French broadcasts appear to be synthesized, but in French only with recorded trailing segment indicating frequency of upcoming transmissions. The transmissions begin with “Origin Meteo France”.
Recording made at 20191113T1621Z_1650kHz.
Recording made at 20191113T1702Z_1650kHz.
Recording made at 20191119T1922Z_2677kHz.
Recording made at 20191119T1937Z_2677kHz.
Greenland
Transmissions from Greenland appear to be voiced by a live person in English, followed by Greenlandic.
Recording made 20191114T1237Z_3125kHz.
Recording made 20191114T1307Z_3125kHz.
Islas Baleares
Transmissions in the Balearic Islands are made by a synthesized voice from Palma radio, preceded by “all ships, all ships, all ships”. English transmissions are made first, with transmissions made in Spanish following.
Recording made 20191118T1005Z_1755kHz.
Recording made 20191119T2336Z_1755kHz.
Italy
I was only able to capture one readable transmission from Italy. The synthesis they use is very primitive with a fixed cadence which makes it seem like the radio is “barking” at you. Transmissions are made in English first, then Italian. Transmissions are preceded by a two blips.
Recording made at 20191114T1418Z_2656kHz.
Monaco
Transmissions from Monaco Radio are easily recognised by music preceding the transmission, with transmission in French followed by English, then closed off with the same music. Towards the end of the transmissions, information about frequencies and other means of obtaining weather information are provided. The voices appear to be pre-recorded.
Recording made at 20191118T1500Z_4363kHz.
Recording made at 20191119T0700Z_4363kHz.
Netherlands
Some transmissions from the Netherlands Coast Guard appear to be made in English using human voices, albeit some pre-recorded, with clear identification at the beginning of the transmission. There are synthesized voice programs as well.
Recording made at 20191101T1133Z_3673kHz. Begins with a friendly “Hello all ships”, reporting no safety messages on hand.
Recording at 20191101T1252Z_3673kHz issuing gale warning after a securite call. Male and female voices.
Recording made at 20191102T0942Z_3673kHz with the North sea shipping forecast, preceded by “all ships”. Transmission made with synthesized voice, with gaps between words.
New Zealand
The transmissions from Taupo Maritime Radio (ZLM) are quite familiar, with a synthesized deep-male voice speaking in English. The voice synthesis system seems to be a hybrid of recorded key-words and synthesized words, with a very staccato rhythm.
Recorded 20191102T0036Z_6224kHz coastal navigation warnings and weather.
Recorded 20191102T0303Z_6224kHz, navigational warnings preceded by “all stations” call.
Norway
Transmissions from the Norweigan Coastal Radio (South) are made with a mixture of live voice and synthesized voice in English followed by Norwegian. The voice synthesis on this system is actually quite smooth-sounding.
Recording at 20191119T2234Z_1665kHz from Norweigan Coastal Radio (South) containing warnings spoken by a live female voice making a securite call to all stations.
Recording at 20191119T2315Z_1665kHz from Norwegian Coastal Radio (South) containing weather by synthesized female voice in English and synthesized male voice in Norwegian. Addressed to “all stations”.
Recording at 20191120T0234Z_1665kHz from Norweigan Coastal Radio (South) containing warnings and safety information spoken by a live male voice making a securite call to all stations.
Recording at 20191120T0634Z_1665kHz from Norwegian Coastal Radio (South) containing maritime safety information by synthesized female voice in English and synthesized male voice in Norwegian.
Poland
Transmissions from Poland were difficult to receive and appear to be transmitted by live voice bilingually.
Recording made at 20191103T1337Z_2720khz. Recording seems to be interfered with, resulting in a jittery quality.
Recording made at 20191103T1936Z_2720kHz.
Spain
Spain’s broadcasts are mostly bilingual using voice synthesis, first in Spanish, followed by English. Transmissions are preceded by a securite call.
Recording at 20191108T1505Z_1677kHz. 
Recording at 20191108T2306Z_1677kHz, with interference from STANAG transmission in the upper frequencies.
Recording at 20191109T0806Z_1677kHz with a live voice, seemingly the exception.
Thailand
Thailand’s transmissions are very familiar to me, as their broadcast slots are always either filled with a melodic tune or the information in either synthesized English or Thai.
Recording made at 20191031T0000Z_8743kHz.
United Kingdom
All of the broadcasts from the UK appear to be voiced by a live person in English.
Recording made 20191029T1933Z_2226kHz from Aberdeen Coast Guard.
Recording made 20191030T0430Z_2226kHz from Aberdeen Coast Guard.
Recording made 20191030T0733Z_2226kHz from Aberdeen Coast Guard.
Recording made 20191109T0835Z_1677khz from Malin Head Coast Guard.
United States of America
Transmissions from the USA are made using a booming synthesized male voice which is quite distinctively recognisable by the level of intonation on the word “feet” at times. Many of the broadcasts can be quite long.
Recording made at 20191119T1805Z_8764kHz.
Recording made at 20191120T0930Z_8764kHz.
Vietnam
Transmissions from Vietnam appear only to be made in Vietnamese using live voices, with several stations appearing to be in operation.
Recording made at Vietnam-20191029T0805Z_8294kHz.
Recording made at 20191029T0905Z_8294kHz.
Recording made at 20191029T1035Z_8294kHz.
Recording made at 20191029T1705Z_8294kHz.
Recording made at 20191030T0035Z_8294kHz.
Other Interesting Recordings
While attempting to collect the above recordings, a few interesting recordings were made as well, often by chance.
This recording was made at 20191102T1212Z_4149kHz. I’m not sure what the message conveyed is, but the frequency is used by the Argentinean Coast Guard service – there is a mention of a frequency at the end, but as I never received any other transmission with the same format on that frequency, I would have to consider this transmission unconfirmed.
Some countries appear to use the distress frequency 2182kHz as the broadcast frequency for their weather warnings. However, while monitoring, it seems that Italy offers a reminder on this frequency just prior to their broadcasts to remind mariners to tune in – e.g:
Collisions can also occur on the broadcasts, resulting in overlapping transmissions. In this case, it seems the transmissions from both Spain and Malin Head, UK have collided at 20191108T2035Z_1677kHz.
Conclusion
After months of work, it was possible to hear the HF weather voice transmissions from 18 different countries, some of which use synthesized voice systems, others which use live/recorded voice and others which use a mixture of the two. It is good to know the services are still alive and to preserve their character, as they may be at risk of discontinuation in the future.
While I was pleasantly surprised by how many different stations I was able to receive, there are still a number of countries on the list which were not captured. The fact they were not captured is probably down to limitations in the receivers used, time of day, propagation, and other factors rather than the station being out of service (even though this is a distinct possibility). I will perhaps try again to capture some of the missing stations in the future and post about them separately.
