Hello Jean-Marie, I have to say that I prefer JTDX but I have no data of my own in respect of comparisons with WSJTx. The JTDX author and a small team of testers run the tests and publish decoding comparison tables - the number of decodes per period on average. JTDX is a clear winner even at the previous 16bit versions. As for 32bit, so far only theoretical data has been published but it shows a strong liklihood of more completed QSOs when a band is saturated with signals and the radio, like ours, can take advantage of the increased sampling. Therefore the UMA addition giving JTDX 32 bit info is, as always, just to provide as complete a picture of what is available to EE product owners and should not be taken as any endorsement on my part. So to answer your question, if you prefer WSJTx then I think that is fine, stay with it.
Despite clarification, there is still some mis-understandings being aired on Facebook. Igor UA3DJY, the JTDX author has tried to explain further:
Sampling initially came from analog to digital signal conversion. SDR have ADC, sound card have ADC.
Regarding AF frequency good sound card can take samples from analog signal where each sample has 24-bit resolution.
Built in transceiver audio codecs are usually ‘poor man’s’ HW with 16-bit resolution as most transceivers have less than 90dB IMD3 dynamic range and 24-bit resolution can be useless there. In some cases 24-bit resolution can bring advantage if levels being not matched well, in such scenario 16-bit audio stream can get intermodulation products while 24-bit stream will provide greater dynamic range capacity for unmatched levels applied to ADC.
Sampling frequency: as per Nyquist-Shannon theorem
en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem it is possible to fully recover signal from samples if sampling frequency is twice greater than signal frequency.
JTDX/WSJT-X operating with 48000 samples per second, so it could be possible to recover 24kHz AF signal from audio stream, at the moment we do not need such bandwidth and perform downsampling in JTDX/WSJT-X to 12000 samples per second. We have been using 0...5000 Hz spectrum out of 6000 Hz available.
This statement is correct concerning rc147_32A: "JTDX is not about sample rate, it' about sample
size»
As WDM KS driver being used in SDR does support 24-bit sample size and JTDX 32A does support 32-bit sample size some software shall convert 24-bit to 32-bit (JTDX RX) and back (JTDX TX). For SDR setup such software is virtual audio cable, for sound card it is operating system responsible for conversion.
From coding point of view 32-bit resolution is easier to implement versus 24-bit, that’s why we have got JTDX 32A.
It is coming of cost of extra CPU cycles, as 3rd party software(or operating system) expanding 24-bit signal to 32-bit resolution (JTDX RX) and back (JTDX TX).
73 Igor
So for our SDRs, which fortunately for us are independent of the OS, provided you set VAC as shown in the UMA you can take full advantage of the new JTDX branch (if you want to of course).
73 de Erik.