rcu_audio: improve audio quality
TI documentation for RAS_Decode appears to be
incorrect. Correcting the parameters passed to
RAS_Decode noticeably improves audio quality.
Change-Id: I584ba7077f1e922f80d4b52aff207d237b3bbc75
diff --git a/rcu_audio/ti-rcu-audio.cc b/rcu_audio/ti-rcu-audio.cc
index e174477..803ce3c 100644
--- a/rcu_audio/ti-rcu-audio.cc
+++ b/rcu_audio/ti-rcu-audio.cc
@@ -96,11 +96,19 @@
*
* The first three bytes of payload are some kind of RAS header.
* RAS_Decode() says the data pointer must include the three bytes
- * of header, but the length must not include those 3 bytes.
- * So the length is decremented by 6+1+1+3 for a total of 11.
+ * of header, but that the length must not include those 3 bytes.
+ *
+ * HOWEVER, the documentation is a filthy liar. RAS_Decode decrements
+ * its inputLength by three to account for the header length.
+ * If we obey the instructions and omit those 3 bytes from the
+ * length, the audio quality is noticeably worse because it starts
+ * throwing away three bytes of audio data.
+ *
+ * So both the pointer and the length passed to RAS_Decode *include*
+ * the header bytes.
*/
data = &ibuf[6 + 1 + 1];
- data_len = ilen - (6 + 1 + 1 + 3);
+ data_len = ilen - (6 + 1 + 1);
if (RAS_Decode(RAS_DECODE_TI_TYPE1, data, data_len, obuf, &olen) == 0) {
rcaudio::AudioSamples samples;
char bdaddr[18];
