Update to latest dft_detect (with floating point bug fixes).

auto_rx/test/test_demod.py

@@ -98,7 +98,7 @@ processing_type = {
         # Decimate to a 24 kHz bandwidth, demodulator, then interpolate back up to 48 kHz.
         'demod' : "| csdr fir_decimate_cc 4 0.005 HAMMING 2>/dev/null | csdr fmdemod_quadri_cf | csdr limit_ff | csdr rational_resampler_ff 2 1 0.005 HAMMING | csdr convert_f_s16 | sox -t raw -r 48k -e signed-integer -b 16 -c 1 - -r 48000 -t wav - highpass 20 2>/dev/null| ",
         # Decode using rs41ecc
-        'decode': "../dft_detect3 2>/dev/null",
+        'decode': "../dft_detect 2>/dev/null",
         # Grep out the line containing the detected sonde type.
         "post_process" : " | grep \:"
     },

scan/dft_detect.c

@@ -547,8 +547,8 @@ static int init_buffers() {
     M -= 1;
     N_DFT <<= 1;
 #endif
-    LOG2N = log(N_DFT)/log(2);
-    //while ((1 << LOG2N) < N_DFT) LOGN++;
+    LOG2N = log(N_DFT)/log(2)+0.1; // 32bit cpu ... intermediate floating-point precision
+    //while ((1 << LOG2N) < N_DFT) LOG2N++;  // better N_DFT = (1 << LOG2N) ...
 
     K = M-NN - delay; // N+K < M