dmAudioRateConverterSetParams(3dm)dmAudioRateConverterSetParams(3dm)NAMEdmAudioRateConverterSetParams - set rate converter parameter values
SYNOPSIS
#include <dmedia/dm_audioutil.h>
#include <dmedia/dm_audio.h>
#include <dmedia/dm_params.h>
DMstatus dmAudioRateConverterSetParams(DMaudiorateconverter handle,
DMparams *params)
PARAMETER
handle DMaudiorateconverter structure, created by
dmAudioRateConverterCreate(3dm).
params List of parameters for specification/query. Null (0) value ok.
RETURNED VALUE
Returns DM_SUCCESS or DM_FAILURE.
DESCRIPTIONdmAudioRateConverterSetParams(3dm) sets state of converter with params.
Unrecognized parameters are ignored.
The set of parameters for params are, defined in dmedia/dm_audioutil.h:
DM_AUDIO_RC_INPUT_RATE
DM_AUDIO_RC_OUTPUT_RATE
Audio Util rate conversion input, output rate parameters (rate given in
Hz). The corresponding values for the rates are doubles. The jitter-
free algorithm supports sampling rate conversion between any two of the
following rates in the set {8000, 11025, 16000, 22050, 32000, 44100,
48000} (sample frames/second or Hertz). Polynomial_order_{1,3}
algorithms support arbitrary input and output rate.
DM_AUDIO_RC_ALGORITHM
DM_AUDIO_RC_JITTER_FREE DM_AUDIO_RC_POLYNOMIAL_ORDER_1
DM_AUDIO_RC_POLYNOMIAL_ORDER_3
Page 1
dmAudioRateConverterSetParams(3dm)dmAudioRateConverterSetParams(3dm)
Currently, the algorithms supported are jitter-free, which is at least
one integer interpolator / integer decimator stage, polynomial-order-1,
which is linear interpolation between 2 proximal samples, and
polynomial-order-3, which is rate convert block by 3rd order polynomial
interpolation between four proximal samples.
DM_AUDIO_RC_JITTER_FREE_STOPBAND_ATTENUATION
DM_AUDIO_RC_JITTER_FREE_STOPBAND_ATTENUATION_78_DB
DM_AUDIO_RC_JITTER_FREE_STOPBAND_ATTENUATION_96_DB
DM_AUDIO_RC_JITTER_FREE_STOPBAND_ATTENUATION_120_DB
The parameter is specific to the jitter free rate converter filter
stopband attenuation. The values are limited to set {78, 96, 120}. The
values correspond to the minimum attenuation level of the out-of-band
frequencies, those frequencies which do not fit in the smaller of the
output or input bandwidths. Higher values correspond to better noise
suppression.
DM_AUDIO_RC_JITTER_FREE_TRANSITION_BANDWIDTH
DM_AUDIO_RC_JITTER_FREE_TRANSITION_BANDWIDTH_1_PERCENT
DM_AUDIO_RC_JITTER_FREE_TRANSITION_BANDWIDTH_10_PERCENT
DM_AUDIO_RC_JITTER_FREE_TRANSITION_BANDWIDTH_20_PERCENT
The parameter is specific to the jitter free rate converter filter
transition bandwidth. The values are limited to set {1, 10, 20}. The
values correspond to the % of the high frequency response of the output
signal lost in the conversion process. Lower percentages correspond to
better high frequency response.
Parameters for query only:
DM_AUDIO_RC_ATOMIC_IN_LENGTH
DM_AUDIO_RC_ATOMIC_OUT_LENGTH
The rate convert algorithm processes blocks of a fixed length determined
by the conversion process parameters. The input and output lengths must
be a multiple of DM_AUDIO_RC_ATOMIC_IN_LENGTH and
DM_AUDIO_RC_ATOMIC_OUT_LENGTH, respectively. Query these parameters on a
configured converter can also be done by using
dmAudioRateConverterGetParams(3dm).
DM_AUDIO_RC_GROUP_DELAY Type DM_TYPE_FLOAT
The group delay measured in output samples.
These sampling rate conversion algorithms use filter operations that
convolve a N past input samples with a filter M-coefficient array to
create N+M-1 output samples. Note that he filtered signal contains more
Page 2
dmAudioRateConverterSetParams(3dm)dmAudioRateConverterSetParams(3dm)
samples than the unfiltered signal.
We use linear phase filters with a constant group delay G = (M-1)/2
samples. In real time operation, convolution simply delays the output by
G samples. In file conversion expecting N samples in and out, the output
signal is offset by G samples from the first sample thus omits the last G
samples from the output file. Multiple conversions compound the delays
and data loss.
SGI audio applications manage the above problems by omitting the first G
samples while computing more than N+G samples. In practice, G is rounded
to the nearest sample and thus sample time alignment is occasionally
ahead or behind one sample. G specifies the number of output samples to
omit.
Query DM_AUDIO_RC_GROUP_DELAY on a configured converter can also be done
by using dmAudioRateConverterGetParams(3dm).
NOTEdmAudioRateConverterSetParams(3dm) calls dmAudioRateConverterReset(3dm)
internally.
Files converted using jitter-free algorithm with decibels set to 96 and
bandwidth set to 10 will sound most excellent. For the highest quality
conversion algorithm, however, you can set decibels to 120 and bandwidth
to 1. By decreasing decibels and increasing bandwidth, you increase the
speed of the rate conversion at a loss of high frequency response.
The following table shows the relationships between the bandwidth
parameter, and the pass band ripple:
bandwidth % final passband
parameter bandwidth ripple
tapered
20 20% +/- 0.05 dB
10 10% +/- 0.05 dB
1 1% +/- 0.0005 dB
SEE ALSOdmAudioRateConverterCreate(3dm), dmAudioRateConvert(3dm),
dmAudioRateConverterGetParams(3dm).
Page 3
