The other half of ClackMo. It takes audio — a file, a recording pasted from Sound Recorder, or the microphone — and reads it back into the text that made it, using real demodulators rather than a lookup of what was encoded.
Pick the protocol, check the parameters match how the signal was sent, then choose a source. Listen decodes continuously from the microphone and scrolls the waterfall as it goes; the other three read a whole recording at once. A file can also be dropped anywhere on the window.
The parameters are the tuning controls: the mark and space frequencies, baud rate and framing have to match the transmitter. What the decoder actually measured appears under Signal, which is the quickest way to tell a mistuned parameter from a bad recording.
Open ClackMo in another window, put the two side by side, press Listen here and Play there. The signal goes out of the speakers and back in through the microphone, and what arrives is decoded from the air rather than passed between the two.
DTMF and MF R1 through a Goertzel tone bank; Bell 202, Bell 103, V.21, V.23, RTTY and the Kansas City Standard through a quadrature correlator and asynchronous framing; AX.25 through a bit-clock recovery loop with NRZI, bit destuffing and a verified frame check sequence; ZX Spectrum, Commodore 64 and SMPTE timecode by edge timing, with their checksums verified; Morse from the envelope of the keyed tone; and PSK31 through a differential detector and the varicode alphabet.
V.22 DQPSK and multi-channel CW are encode-only for now. V.22 needs carrier and timing recovery through a scrambler; multi-channel CW needs the channel tones to be found before they can be read, since the transmitter chooses them per line.
Audio is resampled to 48 kHz and mixed to mono before decoding. A protocol is decoded with the parameters shown, not guessed at — there is no automatic identification.