Table of Contents
Advertisement
Below is a visualisation of a six-channel audio capture. Note the relatively quiet mic signals compared
with the reference. This is typical and allows for loud near-end signals without distortion.
PACKING SPECIFIC SIGNALS
PACKED_PIPELINE_OUTPUT, PACKED_MIC, PACKED_REF all use the same underlying packing
function. They pack 2 channels (mic0/1 or ref/ref_r or pipe0/pipe1) into a single audio channel. It
requires that the output interface, including host processing, be capable of bit-perfect 32b audio . It
packs the two 16kHz samples into three 48kHz samples as follows:
}
top 24b of sample[0] with 8b LSB marker '0x00'
}
top 24b of sample[1] with 8b LSB marker '0x01'
}
the bottom 8b of sample[0], the bottom 8b of sample[1], 0x00, 8b LSB marker '0x02'
The unpacker.py script then looks for 0x00, 0x01, 0x02 in the LSByte to check for a packed
sequence. So inspecting the wav in a hex editor should make it clear when it is captured properly.
It will capture bit-perfect data.
Packing specific signals will not work on a Mac because it only supports 24b audio due to core audio
representing audio using single-precision floating-point. It has been tested and works well on Linux
(x86/RPI) which supports bit-perfect 32b audio.
XM-014232-PC
86
Advertisement
Table of Contents
