Corrected Biquad filter functions and added 24 and 32 bit note to I2S.

pull/16/head
boblark 3 years ago
parent 8e470ff2ce
commit 1ad067459b
  1. 80
      docs/index.html

@ -1537,9 +1537,8 @@ look ahead delay, as well.</p>
<script type="text/x-red" data-help-name="AudioFilterBiquad_F32">
<h3>Summary</h3>
<div class=tooltipinfo>
<p>Biquadratic cascaded filter, useful for all sorts of filtering.
Up to 4 stages may be cascaded.
</p>
<p>Biquadratic cascaded IIR filters, useful for all sorts of
frequency filtering. Up to 4 stages may be cascaded. </p>
<p align=center><img src="img/biquad.png"></p>
</div>
<h3>Audio Connections</h3>
@ -1549,6 +1548,11 @@ look ahead delay, as well.</p>
<tr class=odd><td align=center>Out 0</td><td>Filtered Signal Output</td></tr>
</table>
<h3>Functions</h3>
<p class=func><span class=keyword>begin</span>();</p>
<p class=desc> This <b>required</b> function initializes
the BiQuad instance using the ARM DSP Math Library. There are no
calling parameters.
</p>
<p class=func><span class=keyword>setLowpass</span>(stage, frequency, Q);</p>
<p class=desc>Configure one stage of the filter (0 to 3) with low pass
response, with the specified corner frequency and Q shape. If Q is
@ -1596,9 +1600,7 @@ look ahead delay, as well.</p>
<p class=desc>Configure one stage of the filter (0 to 3) with an arbitrary
filter response. The array of coefficients is in order: B0, B1, B2, A1, A2.
Each coefficient must be less than 2.0 and greater than -2.0. The array
should be type double. Alternately, it may be type int, where 1.0 is
represented with 1073741824 (2<sup>30</sup>).
</p>
should be type double. </p>
<p class=func><span class=keyword><strong>double* </strong>getCoefficients</span>();</p>
<p class=desc>Returns a pointer to the array of double precision coefficients. For
up to four stages, each stage is arranged in order B0, B1, B2, A1, A2. This is a
@ -1607,17 +1609,25 @@ look ahead delay, as well.</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Effects &gt; Filter
</p>
<h3>Notes</h3>
<p>Filters can with gain must have their input signals attenuated, so the
signal does not exceed 1.0.
</p>
<p>This object implements up to 4 cascaded stages. Unconfigured stages will
not pass any signal.
</p>
<p>Biquad filters with low corner frequency (under about 400 Hz) can run into
trouble with limited numerical precision, causing the filter to perform
poorly. For very low corner frequency, the State Variable (Chamberlin)
filter should be used.
</p>
<p>Each instance of the Biquad filter class can have 0 to 4 cascaded biquad filters,
with each independent. These can mix filter types, such as Low Pass and High Pass
or they can be multiples of the same type. Note that, in general, cascading identical
Biquad IIR filters will not be the most useful. Check the Internet for discussions
of cascading filters to achieve specific responses, such as Butterworth
or Chebychev.
</p>
<p>This object implements up to 4 cascaded stages. The four biquads per instance
can each be used, or not used, and the unused
ones will be treated as pass throughs.
</p>
<p>These IIR filters do not provide flat time delay with frequency as provided
by symmetrical FIR filters. If this is important, the extra complexity of the FIR
type may be justified.</p>
<p>These IIR filters generally do not have the degrees of freedom that FIR filters easily provide.
That means the response shapes are generally more constrained. This can be overcome
by adding many more biquad sections, with properly chosen frequencies and Qs, but
then it becomes easier to just use the FIR. Thus, most applications where the Biquad IIR
is appropriate will fit into this 4-filter object.</p>
</script>
<script type="text/x-red" data-template-name="AudioFilterBiquad_F32">
<div class="form-row">
@ -1669,6 +1679,7 @@ look ahead delay, as well.</p>
a non-symetrical response, make sure the data is in reverse time order.
</p>
<h3>Notes</h3>
<p>FIR filters requires more CPU time than Biquad (IIR), but they can
implement filters with better phase response.
</p>
@ -1986,36 +1997,17 @@ look ahead delay, as well.</p>
<p>Audio from
master mode I2S may be used in the same project as ADC, DAC and
PWM signals, because all remain in sync to Teensy's timing</p>
<!--
<h3>Examples</h3>
<p class=exam>File &gt; Examples &gt; Audio &gt; HardwareTesting &gt; PassThroughStereo
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Recorder
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Analysis &gt; PeakMeterStereo
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Analysis &gt; FFT
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Analysis &gt; SpectrumAnalyzerBasic
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Effects &gt; Chorus
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Effects &gt; Flange
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Effects &gt; Filter
</p>
<p class=exam>File &gt; Examples &gt; Audio &gt; Effects &gt; Filter_FIR
</p>
-->
<p Most examples use the Audio card with I2S transfer.</p>
<h3>Notes</h3>
<p>Only one I2S input and one I2S output object may be used. Master
and slave modes may not be mixed (both must be of the same type).
</p>
<p>Being an "_F32" object, audio is passed into this class using F32 memory
that was allocated using an AudioMemory_F32() call in your sketch's setup()
routine. But, internally, this class still uses some Int16 data handling,
so be sure to also include an AudioMemory() call in addition to AudioMemory_F32()
to allocate the Int16 memory.</p>
and slave modes may not be mixed (both must be of the same type).</p>
<p>As of May 2022, the I2S input and output objects can handle 24 or 32-bit
data transfer, if that is supported by the audio hardware device.
No class functions are needed. It is "automagic".</p>
</script>
<script type="text/x-red" data-template-name="AudioInputI2S_F32">
<div class="form-row">

Loading…
Cancel
Save