Figure 13.
STSW-BLUEMIC-1 audio processing chain
1.2.7
ADPCM compression
The ITU-T G.726 adaptive differential pulse code modulation (ADPCM) standard is applied to save bandwidth.
This audio algorithm for lossy waveform coding predicts the current signal value from previous values, and
transmits the difference between the real and the predicted value, quantized with an adaptive quantization step.
The ADPCM algorithm used in this application compresses digital voice signals encoded as:
•
Audio format: PCM
•
Audio sample size: 16 bits
•
Channels: 1 (mono)
•
Audio sample rate: 8-16 kHz
Figure 14.
ADPCM encode-decode schema
BlueVoiceADPCM implements a modified version of the compression algorithm with improved communication
robustness through an additional low data rate channel with collateral information added to the ADPCM quantized
values; slightly increasing the overall bit-rate to an average 64.3 Kbps.
The internal buffering required by ADPCM compression is shown in the figure below. 16-bit input PCM samples
are encoded in 8-bit temporary samples with 4-bit actual data (u8 ADPCM_app buffer) and then encapsulated in
8-bit samples containing information of two PCM samples (u8 ADPCM buffer).
UM2257
Software description
UM2257
-
Rev 2
page 10/35
