Offset off by one in parsescp.c

The title says it all. Encoding type it read from the second byte of sampling period, and compression byte is read from the encoding byte. Here is the patch:

--- convert/parsescp.c.orig	2018-11-28 22:55:57.000000000 +0100
+++ convert/parsescp.c	2019-07-28 21:32:33.361920818 +0200
@@ -1958,8 +1958,8 @@
 	printf("Error: sampling frequency not specified;  assuming 500 Hz\n");
 	sfreq = 500;
     }
-    if (vflag) printf("  Encoding type: %d ", *(p+3));
-    switch (*(p+3)) {
+    if (vflag) printf("  Encoding type: %d ", *(p+4));
+    switch (*(p+4)) {
     case 0: if (vflag) printf("(amplitudes)\n"); enctype = 0; break;
     case 1: if (vflag) printf("(first differences)\n"); enctype = 1; break;
     case 2: if (vflag) printf("(second differences)\n"); enctype = 2; break;
@@ -1968,7 +1968,7 @@
     }
     if (section == 6) {
 	if (vflag) printf("  Compression type: %d ", *(p+4));
-	switch (*(p+4)) {
+	switch (*(p+5)) {
 	case 0: if (vflag) printf("(non-bimodal)\n"); comptype = 0; break;
 	case 1: if (vflag) printf("(bimodal)\n"); comptype = 1; break;
 	default: if (vflag) printf("<undefined, assuming bimodal>\n"); comptype = 1;

Start in convert/parsescp.c around the sampling-period parsing shown in the issue, and trace the byte offsets used for encoding and compression, including section 6. Confirm the diagnostic output and parsed values use the intended bytes for both fields, then validate the affected SCP parsing behavior with representative input.