commit b5bf0294 moved the if(need_new_coefficients) set_passthrough(equ)
after the if(is_passthrough) return FLOW_OK shortcut, so the passthrough
mode would never get updated even if the coefficients change.
Fixes equalizer-test doing .. nothing.
Add private replacements for deprecated functions such as
g_mutex_new(), g_mutex_free(), g_cond_new() etc., mostly
to avoid the deprecation warnings. We'll change these
over to the new API once we depend on glib >= 2.32.
This allows per band configuration of a peak, low shelf or
high shelf filter, which can be very useful if the band frequencies
and widths are manually configured.
Refactor the filter setup. Add two new filters with shelf characteristics for
first and last band. Change gain calculation as recommended in the quoted
document (no qrt needed). Rename variables to match the formulas in the
document.
Move reallocating the history buffer out of _compute_frequencies() and call the
right function as needed. Add some logging and tweak the formatting of existing
logging. Simplify setting need_new_coefficients when changing properties.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c: (update_coefficients):
Don't calculate the filter coefficients for every single buffer
but only when it's needed. Fixes bug #557260.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c:
(gst_iir_equalizer_transform_ip):
Don't clip float/double samples, correctly unset passthrough mode
and use better rounding for integer samples.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c:
(gst_iir_equalizer_band_set_property), (gst_iir_equalizer_init),
(setup_filter), (set_passthrough), (update_coefficients),
(gst_iir_equalizer_compute_frequencies),
(gst_iir_equalizer_transform_ip):
* gst/equalizer/gstiirequalizer.h:
Update the filter coefficients only when needed in the transform_ip
function and correctly set the element into passthrough mode if the
gain of all bands is 0.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c:
Use a bigger type in integer mode for the intermediate results to
prevent overflows. This fixes the crippled sound when using the
equalizer in integer mode. Fixes bug #510865.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c: (gst_iir_equalizer_finalize):
Unparent all bands from the equalizer when finalizing to stop
leaking them.
Original commit message from CVS:
* gst/equalizer/demo.c: (main):
* gst/equalizer/gstiirequalizer.c:
(gst_iir_equalizer_band_class_init), (setup_filter):
Allow setting 0 as bandwidth and handle this correctly.
Also handle a bandwidth of rate/2 properly.
* gst/equalizer/gstiirequalizernbands.c:
(gst_iir_equalizer_nbands_class_init):
Make it possible to generate a N-band equalizer with 1 bands. The
previous limit of 2 was caused by a nowadays replaced calculation
doing a division by zero if number of bands was 1.
Original commit message from CVS:
* gst/equalizer/Makefile.am:
* gst/equalizer/demo.c: (on_window_destroy), (on_configure_event),
(on_gain_changed), (on_bandwidth_changed), (on_freq_changed),
(draw_spectrum), (message_handler), (main):
Add small demo application based on the spectrum demo applications
that gets white noise as input, pushes it through an equalizer and
paints the spectrum. For every equalizer band it's possible to set
gain, bandwidth and frequency.
* gst/equalizer/gstiirequalizer.c: (setup_filter):
Add some guarding against too large or too small frequencies and
bandwidths. Also improve debugging a bit.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c:
(gst_iir_equalizer_band_set_property),
(gst_iir_equalizer_band_get_property),
(gst_iir_equalizer_band_class_init), (arg_to_scale),
(setup_filter), (gst_iir_equalizer_compute_frequencies):
Replace filters with a bit better filters for which we can actually
find documentation, which don't change anything on zero gain, etc.
Make the frequency property of the bands writable, rename the
band-width property to bandwidth and change the meaning to the
frequency difference between bandedges, change the meaning of the
gain property to dB instead of a weird scale between -1 and 1 that
has no real meaning.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c:
(gst_iir_equalizer_band_set_property),
(gst_iir_equalizer_band_get_property),
(gst_iir_equalizer_band_class_init), (gst_iir_equalizer_band_init),
(gst_iir_equalizer_band_get_type), (gst_iir_equalizer_class_init),
(setup_filter), (gst_iir_equalizer_setup):
* gst/equalizer/gstiirequalizer.h:
Move bandwidth property to the separate bands and add float64 support.
Original commit message from CVS:
* gst/equalizer/gstiirequalizer.c:
(gst_iir_equalizer_band_set_property),
(gst_iir_equalizer_child_proxy_get_child_by_index),
(gst_iir_equalizer_child_proxy_get_children_count),
(gst_iir_equalizer_child_proxy_interface_init),
(gst_iir_equalizer_class_init), (arg_to_scale), (setup_filter),
(gst_iir_equalizer_compute_frequencies):
* gst/equalizer/gstiirequalizer10bands.c:
(gst_iir_equalizer_10bands_class_init):
* gst/equalizer/gstiirequalizer3bands.c:
(gst_iir_equalizer_3bands_class_init):
* gst/equalizer/gstiirequalizernbands.c:
Better algorith for the center frequencies. Subtract band filters from
input for negative gains. Rework the gain mapping.
