decoder/wavpack: read float samples as-is, don't convert to integer

For MPD's new floating point support: when a decoded wavpack file needs to be resampled, don't convert float to int and back to float.

decoder/wavpack: read float samples as-is, don't convert to integer
92bb10ee · Max Kellermann · 8465c5fe · 92bb10ee
Commit 92bb10ee authored Oct 08, 2011 by Max Kellermann
Hide whitespace changes
Inline Side-by-side

Showing with 4 additions and 5 deletions

wavpack_decoder_plugin.c src/decoder/wavpack_decoder_plugin.c +4 -5

No files found.
--- a/src/decoder/wavpack_decoder_plugin.c
+++ b/src/decoder/wavpack_decoder_plugin.c
@@ -111,12 +111,11 @@ static void
 format_samples_float(G_GNUC_UNUSED int bytes_per_sample, void *buffer,
 		     uint32_t count)
 {
-	int32_t *dst = buffer;
+	float *p = buffer;
-	float *src = buffer;
-	assert_static(sizeof(*dst) <= sizeof(*src));
 	while (count--) {
-		*dst++ = (int32_t)(*src++ + 0.5f);
+		*p /= (1 << 23);
+		++p;
 	}
 }
@@ -127,7 +126,7 @@ static enum sample_format
 wavpack_bits_to_sample_format(bool is_float, int bytes_per_sample)
 {
 	if (is_float)
-		return SAMPLE_FORMAT_S24_P32;
+		return SAMPLE_FORMAT_FLOAT;
 	switch (bytes_per_sample) {
 	case 1: