This block performs CVSD audio encoding. Its design and implementation is modeled after the CVSD encoder/decoder specifications defined in the Bluetooth standard.
More...
|
| ~vocoder_cvsd_encode_sb () |
|
short | min_step () |
|
short | max_step () |
|
double | step_decay () |
|
double | accum_decay () |
|
int | K () |
|
int | J () |
|
short | pos_accum_max () |
|
short | neg_accum_max () |
|
int | work (int noutput_items, gr_vector_const_void_star &input_items, gr_vector_void_star &output_items) |
| just like gr_block::general_work, only this arranges to call consume_each for you More...
|
|
unsigned | decimation () const |
|
void | set_decimation (unsigned decimation) |
|
void | forecast (int noutput_items, gr_vector_int &ninput_items_required) |
| Estimate input requirements given output request. More...
|
|
int | general_work (int noutput_items, gr_vector_int &ninput_items, gr_vector_const_void_star &input_items, gr_vector_void_star &output_items) |
| compute output items from input items More...
|
|
int | fixed_rate_ninput_to_noutput (int ninput) |
| Given ninput samples, return number of output samples that will be produced. N.B. this is only defined if fixed_rate returns true. Generally speaking, you don't need to override this. More...
|
|
int | fixed_rate_noutput_to_ninput (int noutput) |
| Given noutput samples, return number of input samples required to produce noutput. N.B. this is only defined if fixed_rate returns true. Generally speaking, you don't need to override this. More...
|
|
This block performs CVSD audio encoding. Its design and implementation is modeled after the CVSD encoder/decoder specifications defined in the Bluetooth standard.
CVSD is a method for encoding speech that seeks to reduce the bandwidth required for digital voice transmission. CVSD takes advantage of strong correlation between samples, quantizing the difference in amplitude between two consecutive samples. This difference requires fewer quantization levels as compared to other methods that quantize the actual amplitude level, reducing the bandwidth. CVSD employs a two level quantizer (one bit) and an adaptive algorithm that allows for continuous step size adjustment.
The coder can represent low amplitude signals with accuracy without sacrificing performance on large amplitude signals, a trade off that occurs in some non-adaptive modulations.
The CVSD encoder effectively provides 8-to-1 compression. More specifically, each incoming audio sample is compared to an internal reference value. If the input is greater or equal to the reference, the encoder outputs a "1" bit. If the input is less than the reference, the encoder outputs a "0" bit. The reference value is then updated accordingly based on the frequency of outputted "1" or "0" bits. By grouping 8 outputs bits together, the encoder essentially produce one output byte for every 8 input audio samples.
This encoder requires that input audio samples are 2-byte short signed integers. The result bandwidth conversion, therefore, is 16 input bytes of raw audio data to 1 output byte of encoded audio data.
The CVSD encoder module must be prefixed by an up-converter to over-sample the audio data prior to encoding. The Bluetooth standard specifically calls for a 1-to-8 interpolating up-converter. While this reduces the overall compression of the codec, this is required so that the encoder can accurately compute the slope between adjacent audio samples and correctly update its internal reference value.
References:
- Continuously Variable Slope Delta Modulation (CVSD) A Tutorial, Available: http://www.eetkorea.com/ARTICLES/2003AUG/A/2003AUG29_NTEK_RFD_AN02.PDF.
- Specification of The Bluetooth System Available: http://grouper.ieee.org/groups/802/15/Bluetooth/core_10_b.pdf.
- McGarrity, S., Bluetooth Full Duplex Voice and Data Transmission. 2002. Bluetooth Voice Simulink® Model, Available: http://www.mathworks.com/company/newsletters/digest/nov01/bluetooth.html