Fix 8->16 bit output conversion overflow error

The I2S outputs were converting samples in the buffer passed in
from u8->s16 on every call.  When the I2S was full, however, the
sample would not get pushed and the output function would be
called at a later time with the same buffer...which was already
at 16 bits.  It'd be converted again and turned into a high
frequency noise on the output.

Now only modify a copy of the input sample in I2S and I2SNoDAC.
Fixes the conversion problem competely.

Fixes #2
This commit is contained in:
Earle F. Philhower, III committed 2018-05-23 22:40:29 -07:00
1 parent b9be5c69dd
commit 0987b12e58
3 files changed
+17 -10

No files matched your search

+5 -2
View File
@@ -80,11 +80,14 @@ void AudioOutputI2SNoDAC::DeltaSigma(int16_t sample[2], uint32_t dsBuff[8])
bool AudioOutputI2SNoDAC::ConsumeSample(int16_t sample[2])
{
MakeSampleStereo16( sample );
int16_t ms[2];
ms[0] = sample[0];
ms[1] = sample[1];
MakeSampleStereo16( ms );
// Make delta-sigma filled buffer
uint32_t dsBuff[8];
DeltaSigma(sample, dsBuff);
DeltaSigma(ms, dsBuff);
// Either send complete pulse stream or nothing
#ifdef ESP32