.TH DSP 1 dsp\-2.0
.SH NAME
dsp \- an audio processing program with an interactive mode
.SH SYNOPSIS
.B dsp
[\fIoptions\fR] \fIpath\fR ... [\fIeffect\fR [\fIargs\fR]] ...
.SH DESCRIPTION
.B dsp
is an audio processing program with an interactive mode.
.SH OPTIONS
.SS Global options
.TP
\fB\-h\fR
Show help text.
.TP
\fB\-b\fR \fIframes\fR
Block size (must be given before the first input).
.TP
\fB\-i\fR
Force interactive mode.
.TP
\fB\-I\fR
Disable interactive mode.
.TP
\fB\-q\fR
Disable progress display.
.TP
\fB\-s\fR
Silent mode.
.TP
\fB\-v\fR
Verbose mode.
.TP
\fB\-d\fR
Force dithering.
.TP
\fB\-D\fR
Disable dithering.
.TP
\fB\-E\fR
Don't drain effects chain before rebuilding.
.TP
\fB\-p\fR
Plot effects chain magnitude response instead of processing audio.
.TP
\fB\-P\fR
Same as \fB\-p\fR, but also plot phase response.
.TP
\fB\-V\fR
Verbose progress display.
.TP
\fB\-S\fR
Use `sequence' input combining mode.
.SS Input/output options
.TP
\fB\-o\fR
Output.
.TP
\fB\-t\fR \fItype\fR
Type.
.TP
\fB\-e\fR \fIencoding\fR
Encoding.
.TP
\fB\-B\fR/\fBL\fR/\fBN\fR
Big/little/native endian.
.TP
\fB\-r\fR \fIfrequency\fR[\fBk\fR]
Sample rate.
.TP
\fB\-c\fR \fIchannels\fR
Number of channels.
.TP
\fB\-R\fR \fIratio\fR
Buffer ratio.
.TP
\fB\-n\fR
Equivalent to
.EX
	-t null null.
.EE
.SH INPUTS AND OUTPUTS
For a complete list of supported input/output types, run
.EX
	$ dsp -h
.EE
.SS Input combining modes
In concatenate mode (the default), the inputs are concatenated in the order
given and sent to the output. All inputs must have the same sample rate and
number of channels.
.PP
In sequence mode, the inputs are sent serially to the output like concatenate
mode, but the inputs do not need to have the same sample rate or number of
channels. The effects chain and/or output will be rebuilt/reopened when
required. Note that if the output is a file, the file will be truncated if it
is reopened. This mode is most useful when the output is an audio device, but
can also be used to concatenate inputs with different sample rates and/or
numbers of channels into a single output file when used with the \fBresample\fR
and/or \fBremix\fR effects.
.SS Signal generator
The \fBsgen\fR input type is a basic (for now, at least) signal generator that can
generate impulses and exponential sine sweeps. The syntax for the \fIpath\fR
argument is as follows:
.PP
.RS
[\fItype\fR[@\fIchannel_selector\fR][:\fIarg\fR[=\fIvalue\fR]...]][/\fItype\fR...][+\fIlen\fR[\fBs\fR|\fBm\fR|\fBS\fR]]
.RE
.PP
\fItype\fR may be `sine' for sine sweeps or tones, or `delta' for a delta function
(impulse). `sine' accepts the following arguments:
.TP
\fBfreq\fR=\fIf0\fR[\fBk\fR][\-\fIf1\fR[\fBk\fR]]
Frequency. If \fIlen\fR is set and \fIf1\fR is given, an exponential sine sweep
is generated.
.PP
The arguments for `delta' are:
.TP
\fBoffset\fR=\fItime\fR[\fBs\fR|\fBm\fR|\fBS\fR]
Offset in seconds, miliseconds or samples.
.PP
Example:
.EX
	$ dsp -t sgen -c 2 sine@0:freq=500-1k/sine@1:freq=300-800+2 gain -10
.EE
.SH EFFECTS
.SS Complete effects list
.TP
\fBlowpass_1\fR \fIf0\fR[\fBk\fR]
First-order lowpass filter.
.TP
\fBhighpass_1\fR \fIf0\fR[\fBk\fR]
First-order highpass filter.
.TP
\fBallpass_1\fR \fIf0\fR[\fBk\fR]
First-order allpass filter.
.TP
\fBlowshelf_1\fR \fIf0\fR[\fBk\fR] \fIgain\fR
First-order lowshelf filter.
.TP
\fBhighshelf_1\fR \fIf0\fR[\fBk\fR] \fIgain\fR
First-order highshelf filter.
.TP
\fBlowpass_1p\fR \fIf0\fR[\fBk\fR]
Single pole lowpass (EWMA) filter.
.TP
\fBlowpass\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR]
Second-order lowpass filter.
.TP
\fBhighpass\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR]
Second-order highpass filter.
.TP
\fBbandpass_skirt\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR]
Second-order bandpass filter with constant skirt gain.
.TP
\fBbandpass_peak\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR]
Second-order bandpass filter with constant peak gain.
.TP
\fBnotch\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR]
Second-order notch filter.
.TP
\fBallpass\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR]
Second-order allpass filter.
.TP
\fBeq\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBo\fR|\fBh\fR|\fBk\fR] \fIgain\fR
Second-order peaking filter.
.TP
\fBlowshelf\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBs\fR|\fBd\fR|\fBo\fR|\fBh\fR|\fBk\fR] \fIgain\fR
Second-order lowshelf filter.
.TP
\fBhighshelf\fR \fIf0\fR[\fBk\fR] \fIwidth\fR[\fBq\fR|\fBs\fR|\fBd\fR|\fBo\fR|\fBh\fR|\fBk\fR] \fIgain\fR
Second-order highshelf filter.
.TP
\fBlowpass_transform\fR \fIfz\fR[\fBk\fR] \fIwidth_z\fR[\fBq\fR] \fIfp\fR[\fBk\fR] \fIwidth_p\fR[\fBq\fR]
Second-order lowpass transformation filter. Cancels the poles defined by
\fIfz\fR and \fIwidth_z\fR and replaces them with new poles defined by \fIfp\fR and
\fIwidth_p\fR. Gain is unity at DC.
.TP
\fBhighpass_transform\fR \fIfz\fR[\fBk\fR] \fIwidth_z\fR[\fBq\fR] \fIfp\fR[\fBk\fR] \fIwidth_p\fR[\fBq\fR]
Second-order highpass transformation filter. Also known as a Linkwitz
transform (see http://www.linkwitzlab.com/filters.htm#9). Same as
\fBlowpass_transform\fR except the gain is unity at Fs/2.
.TP
\fBlinkwitz_transform\fR \fIfz\fR[\fBk\fR] \fIwidth_z\fR[\fBq\fR] \fIfp\fR[\fBk\fR] \fIwidth_p\fR[\fBq\fR]
Alias for \fBhighpass_transform\fR.
.TP
\fBdeemph\fR
Compact Disc de-emphasis filter.
.TP
\fBbiquad\fR \fIb0\fR \fIb1\fR \fIb2\fR \fIa0\fR \fIa1\fR \fIa2\fR
Biquad filter.
.TP
\fBgain\fR \fIgain_dB\fR
Gain adjustment in decibels.
.TP
\fBmult\fR \fImultiplier\fR
Multiplies each sample by \fImultiplier\fR.
.TP
\fBadd\fR \fIvalue\fR
Applies a DC shift.
.TP
\fBcrossfeed\fR \fIf0\fR[\fBk\fR] \fIseparation\fR
Simple crossfeed for headphones. Very similar to Linkwitz/Meier/CMoy/bs2b
crossfeed.
.TP
\fBmatrix4\fR [\fIoptions\fR] [\fIsurround_level\fR]
2-to-4 channel (2 front and 2 surround) active matrix upmixer designed for
plain (i.e. unencoded) stereo material.
.sp 0.5
The intended speaker configuration is fronts at ±30° and surrounds between
±60° and ±120°. The surround speakers must be calibrated correctly in
level and frequency response for best results. The surrounds should be
delayed by about 10\-25ms (acoustically) relative to the fronts. No
frequency contouring is done internally, so applying low pass and/or
shelving filters to the surround outputs is recommended:
.EX
	matrix4 surround_delay=15m -6 :2,3 lowpass_1 10k :
.EE
The settings shown above (\-6dB surround level, 15ms delay, and 10kHz
rolloff) are a good starting point, but may be adjusted to taste. The
default \fIsurround_level\fR is \-6dB. Applying the \fBdecorrelate\fR effect to the
surround outputs (optionally with the \fB\-m\fR flag) seems to further improve
the spatial impression (note: adjust `surround_delay' to compensate for
the \fBdecorrelate\fR effect's group delay).
.sp 0.5
The front outputs replace the original input channels and the surround
outputs are appended to the end of the channel list.
.sp 0.5
Options are given as a comma-separated list. Recognized options are:
.RS
.TP
no_dir_boost
Disable directional boost of front channels.
.TP
show_status
Show a status line (slightly broken currently, but still useful for
debugging).
.TP
signal
Toggle the effect when `effect.signal()' is called.
.TP
linear_phase (\fBmatrix4_mb\fR only)
Apply an FIR filter to correct the phase distortion caused by the IIR
filter bank. Has no effect with \fBmatrix4\fR. Requires the \fBfir\fR effect.
.TP
surround_delay=\fIdelay\fR[\fBs\fR|\fBm\fR|\fBS\fR]
Surround output delay. Default is zero.
.TP
filter_type=\fIfilter\fR[:\fIstop_dB\fR[:\fIstop_dB\fR]] (\fBmatrix4_mb\fR only)
Type of filter used for low pass sections of the filter bank. \fIfilter\fR
may be \fIbutterworth\fR, \fIchebyshev1\fR, \fIchebyshev2\fR, or \fIelliptic\fR
(default).
.sp 0.5
The optional \fIstop_dB\fR parameter(s) set the stopband attenuation in
decibels for the Chebyshev and elliptic filters. Only the first
parameter is used for \fIchebyshev1\fR and \fIchebyshev2\fR. For \fIelliptic\fR,
the first parameter applies to the lowpass and the second to the
highpass. If only one parameter is given, it applies to both stopbands.
Default values are 25 for \fIchebyshev1\fR and \fIchebyshev2\fR, and 35:50 for
\fIelliptic\fR.
.RE
.TP
\fBmatrix4_mb\fR [\fIoptions\fR] [\fIsurround_level\fR]
Like the \fBmatrix4\fR effect, but divides the input into eleven individually
steered bands in order to improve separation of concurrent sound sources.
See the \fBmatrix4\fR effect description for more information.
.TP
\fBremix\fR \fIselector\fR|. ...
Select and mix input channels into output channels. Each selector argument
specifies the input channels to be mixed to produce an output channel. `.'
selects no input channels. For example,
.EX
	remix 0,1 2,3
.EE
mixes input channels 0 and 1 into output channel 0, and input channels 2
and 3 into output channel 1.
.EX
	remix -
.EE
mixes all input channels into a single output channel. The active channel
selector is used as an input channel mask for the selector arguments.
.TP
\fBst2ms\fR
Convert stereo to mid/side.
.TP
\fBms2st\fR
Convert mid/side to stereo.
.TP
\fBdelay\fR [\fB\-f\fR [\fIorder\fR]] \fIdelay\fR[\fBs\fR|\fBm\fR|\fBS\fR]
Delay line. The unit for the \fIdelay\fR argument depends on the suffix used:
`\fBs\fR' is seconds (the default), `\fBm\fR' is milliseconds, and `\fBS\fR' is samples. If
\fIdelay\fR is negative, a positive delay is applied to all channels which are
\fBnot\fR selected (except when plotting\(eman actual negative delay is
possible in that case).
.sp 0.5
By default, the delay is rounded to whole samples. The \fB\-f\fR option enables
fractional delay using Thiran allpass interpolation. The \fIorder\fR argument
sets the allpass filter order and may be any integer from 1 through 50. The
default value is 5.
.TP
\fBresample\fR [\fIbandwidth\fR] \fIfs\fR[\fBk\fR]
Sinc resampler. Ignores the channel selector.
.TP
\fBfir\fR [\fIinput_options\fR] [file:][~/]\fIfilter_path\fR|coefs:\fIlist\fR[/\fIlist\fR...]
Non-partitioned 64-bit direct or FFT convolution. Latency is zero for
filters up to 16 taps. For longer filters, the latency is equal to the
\fIfft_len\fR reported in verbose mode. Each \fIlist\fR is a comma-separated list
of coefficients for one filter channel. Missing values are filled with
zeros.
.sp 0.5
The \fIinput_options\fR are useful mostly when loading raw (headerless) input
files and are as follows:
.RS
.TP
\fB\-t\fR \fItype\fR
Type.
.TP
\fB\-e\fR \fIencoding\fR
Encoding.
.TP
\fB\-B\fR/\fBL\fR/\fBN\fR
Big/little/native endian.
.TP
\fB\-r\fR \fIfrequency\fR[\fBk\fR]
Sample rate.
.TP
\fB\-c\fR \fIchannels\fR
Number of channels.
.RE
.TP
\ 
By default, the sample rate of the filter must match that of the effect.
Mismatches may be ignored by setting the sample rate to ``any''.
.TP
\fBfir_p\fR [\fIinput_options\fR] [\fImax_part_len\fR] [file:][~/]\fIfilter_path\fR|coefs:\fIlist\fR[/\fIlist\fR...]
Zero-latency non-uniform partitioned 64-bit direct/FFT convolution. Usually
a bit slower than the \fBzita_convolver\fR effect except for very long filters
on some hardware. \fImax_part_len\fR must be a power of 2 and has a default
value of 16384. Each \fIlist\fR is a comma-separated list of coefficients for
one filter channel. Missing values are filled with zeros.
.sp 0.5
See the \fBfir\fR effect description for an explanation of the \fIinput_options\fR.
.TP
\fBzita_convolver\fR [\fIinput_options\fR] [\fImin_part_len\fR [\fImax_part_len\fR]] [file:][~/]\fIfilter_path\fR|coefs:\fIlist\fR[/\fIlist\fR...]
Partitioned 32-bit FFT convolution using the zita-convolver library.
Latency is equal to \fImin_part_len\fR (64 samples by default).
\fI{min,max}_part_len\fR must be powers of 2 between 64 and 8192. Each \fIlist\fR
is a comma-separated list of coefficients for one filter channel. Missing
values are filled with zeros.
.sp 0.5
See the \fBfir\fR effect description for an explanation of the \fIinput_options\fR.
.TP
\fBhilbert\fR [\fB\-pz\fR] [\fB\-a\fR \fIangle\fR] \fItaps\fR
Simple FIR approximation of a Hilbert transform. The number of taps must be
odd. Bandwidth is controlled by the number of taps. If \fB\-p\fR is given, the
\fBfir_p\fR convolution engine is used instead of the default \fBfir\fR engine.
Similarly, if \fB\-z\fR is given, \fBzita_convolver\fR is used (if available).
The \fB\-a\fR option sets the phase shift in degrees. The default is -90°.
.TP
\fBdecorrelate\fR [\fB\-m\fR] [\fB\-s\fR \fIseed\fR] [\fIstages\fR]
Allpass decorrelator as described in ``Frequency-Dependent Schroeder
Allpass Filters'' by Sebastian J. Schlecht (doi:10.3390/app10010187).
If \fB\-m\fR is given, the same filter parameters are used for all input
channels. The default number of stages is 5, which results in an
average group delay of about 9.5ms at high frequencies. The \fB\-s\fR option
sets the random seed for filter parameter generation.
.TP
\fBnoise\fR \fIlevel\fR[\fBb\fR]
Add TPDF noise. The \fIlevel\fR argument specifies the peak level of the noise
in dBFS if no suffix is given, or the effective precision in bits if the
`\fBb\fR' suffix is given.
.TP
\fBdither\fR [\fIshape\fR] [[\fIquantize_bits\fR] \fIbits\fR]
Apply dither with optional noise shaping. The \fIshape\fR argument determines
the type of dither and the noise shaping filter (if any):

.TS
tab (|);
lB lB
lI l.
Name|Description
_
flat|Flat TPDF with no feedback (default).
sloped|Flat TPDF with feedback. First-order highpass response.
sloped2|Sloped TPDF with feedback. Stronger HF emphasis than \fIsloped\fR.
lipshitz|5-tap E-weighted curve from [1]. Notches around 4k and 12k.
wan3|3-tap F-weighted curve from [2]. Notch around 4k.
wan9|9-tap F-weighted curve from [2]. Notches around 3.5k and 12k.
.TE

The \fIbits\fR argument sets the dither level in bits. The \fIquantize_bits\fR
argument sets the number of levels to quantize to. The default setting for
both is `auto'. If \fIbits\fR is not `auto', dither is applied at the specified
bit depth regardless of the output sample format. \fIbits\fR may be any number.
\fIquantize_bits\fR must be an integer between 2 and 32. If \fIquantize_bits\fR is
not given, it is set to the same value as \fIbits\fR (rounded to the nearest
integer).
.sp 0.5
Note: Currently, setting \fIbits\fR to \fIauto\fR disables dither if the effect
is loaded via \fBwatch\fR or used in \fBladspa_dsp\fR.
.sp 0.5
.RS
.IP [1] 4
S. P. Lipshitz, J. Vanderkooy, and R. A. Wannamaker,
``Minimally Audible Noise Shaping,'' J. AES, vol. 39, no. 11,
November 1991
.IP [2] 4
R. A. Wannamaker, "Psychoacoustically Optimal Noise Shaping,"
J. AES, vol. 40, no. 7/8, July 1992
.RE
.TP
\fBladspa_host\fR [~/]\fImodule_path\fR \fIplugin_label\fR [\fIcontrol\fR ...]
Apply a LADSPA plugin. Supports any number of input/output ports (with
the exception of zero output ports). If a plugin has one or zero input
ports, it will be instantiated multiple times to handle multi-channel
input.
.sp 0.5
Controls which are not explicitly set or are set to `-' will use default
values (if available).
.sp 0.5
The `LADSPA_PATH' environment variable can be used to set the search path
for plugins.
.TP
\fBstats\fR [\fIref_level\fR]
Display the DC offset, minimum, maximum, peak level (dBFS), RMS level
(dBFS), crest factor (dB), peak count, peak sample, number of samples, and
length (s) for each channel. If \fIref_level\fR is given, peak and RMS levels
relative to \fIref_level\fR will be shown as well (dBr).
.TP
\fBwatch\fR [\fB\-e\fR] [~/]\fIpath\fR
Load effects from a file into a sub-chain and reload if the file is
modified. Other than the automatic reload, the behavior is similar to
sourcing a file using the `@' directive (see ``Effects Files''). Some
restrictions apply to automatic reload:
.RS
.IP \(bu 3
The new sub-chain must have the same output sample rate and number of
channels as the previous sub-chain.
.IP \(bu 3
The new sub-chain must not require larger buffers than the previous
sub-chain.
.RE
.TP
\ 
If these conditions are not met, the new sub-chain will not be applied and
an error message will be printed.
.sp 0.5
Currently, this effect polls for file modifications once per second.
Support `inotify` events my be added in the future. Ideally, file
modifications should be atomic (i.e. by writing to a temporary file, then
\fIrename\fR(3)-ing it over top of the original file). If this is not possible,
the \fB\-e\fR flag may be given, which enforces an end-of-file marker in order
to detect partially-written files. This marker, `#EOF#', must be placed at
the beginning of a line and may only be followed by whitespace characters.
.SS Selector syntax
.TS
tab (|);
lB lB
lB l.
Example|Description
_
<empty>|all
\-|all
2\-|2 to n
\-4|0 through 4
1,3|1 and 3
1\-4,7,9\-|1 through 4, 7, and 9 to n
.TE
.PP
Note: There is no difference between `1,3' and `3,1'. Order is not
preserved.
.SS Filter width
The following suffixes are supported:

.TS
tab (|);
lB lB
lB l.
Suffix|Description
_
q|Q-factor (default).
s|Slope (shelving filters only).
d|Slope in dB/octave (shelving filters only).
o|Bandwidth in octaves.
h|Bandwidth in Hz.
k|Bandwidth in kHz.
.TE
.PP
Note: The `\fBd\fR' width suffix also changes the definition of \fIf0\fR from center
frequency to corner frequency (like Room EQ Wizard and the Behringer DCX2496).
.PP
Additionally, a macro is provided for constructing arbitrary-order Butterworth
filters from cascaded second-order sections: `bw<\fIorder\fR>[.\fIn\fR]', where `<\fIorder\fR>' is
the filter order and `\fIn\fR' is an index corresponding to a particular pair of
poles. The Q-factors are always in ascending order. For example,
.EX
	lowpass 1k bw6.0 lowpass 1k bw6.1 lowpass 1k bw6.2
.EE
creates a 6th-order Butterworth lowpass filter. Odd-order filters require an
additional first-order section:
.EX
	lowpass_1 1k lowpass 1k bw5.0 lowpass 1k bw5.1
.EE
.SS File paths
On the command line, relative paths are relative to `$PWD'. Within an effects
file, relative paths are relative to the directory containing said effects
file. A `~/' prefix will be expanded to the contents of `$HOME'. The following
substitutions are supported anywhere within a file path:

.TS
tab (|);
lB lB
l l.
Sequence|Substitution
_
%r|Sample rate in Hz
%k|Sample rate in kHz
%c|Number of channels
%%|Literal `%'
.TE
.SS Channel selectors and masks
A colon (`:') followed by a selector (see ``Selector syntax'') specifies the
input channels for effects that follow. For example,
.EX
	:0,2 eq 1k 1.0 -6
.EE
will apply an \fBeq\fR effect to channels 0 and 2. If an effect changes the total
number of channels, the last channel selector given is parsed again. Additional
channels are not added unless the selector includes an unbounded range.
.PP
Channel numbers refer to the channels in the active channel mask, which is a
property of the containing block. Blocks may be created using braces
(`{ ... }') or by sourcing a file (see ``Effects files''). The channel mask is
derived from the active channel selector at creation. For example,
.EX
	:1,3 { :0 gain -6 :1 gain +6 }
.EE
creates a block with the mask `1,3'. Within the block, `:0' selects the first
channel in the mask (channel 1), and `:1' selects the second channel in the
mask (channel 3). Channel selectors have block scope.
.PP
Channels are automatically added or removed from the active channel mask if an
effect changes the total number of channels. Additional channels are always
appended to the end of the channel list.
.SS Effects files
Files may be sourced using the `@' directive:
.EX
	@[~/]path/to/file
.EE
See ``File paths'' for more information about how paths are interpreted. Note that sourcing
a file implicitly creates a block (see ``Channel selectors and masks''). Within a
file, lines in which the first non-whitespace character is `#' are ignored. A
backslash (`\\') may be used to escape whitespace, `#', or `\\'. Example:
.EX
	gain -4.0
	# This is a comment
	lowshelf 90 1s +4 eq 3k 1.5 -3
.EE
.SS Other directives
An exclamation mark (`!') allows initialization failure of the effect that
follows.
.SS FFTW wisdom
Effects utilizing FFTW3 can optionally load and save wisdom. For \fBdsp\fR, set
`$DSP_FFTW_WISDOM_PATH'. \fBladspa_dsp\fR uses `$LADSPA_DSP_FFTW_WISDOM_PATH'
instead. If a path is set, FFTW plans are created with the FFTW_MEASURE flag.
Accumulated wisdom is written on exit.
.SH EXAMPLES
Read \fIfile.flac\fR, apply a bass boost, and write to alsa device \fIhw:2\fR:
.EX
	dsp file.flac -ot alsa -e s24_3 hw:2 lowshelf 60 0.5 +4
.EE
.PP
Plot the magnitude vs frequency response of an effects chain:
.EX
	dsp -pn [effect [args]] ... | gnuplot
.EE
.PP
Implement an LR4 crossover at 2.2KHz, where output channels 0 and 1 are the
left and right tweeters, and channels 2 and 3 are the left and right woofers,
respectively:
.EX
	dsp stereo_file.flac -ot alsa -e s32 hw:3 remix 0 1 0 1
	  :0,1 highpass 2.2k 0.7071 highpass 2.2k 0.7071 :
	  :2,3 lowpass 2.2k 0.7071 lowpass 2.2k 0.7071 :
.EE
.PP
Apply effects from a file:
.EX
	dsp file.flac @eq.txt
.EE
.SH LADSPA FRONTEND
.SS Configuration
\fBladspa_dsp\fR looks for configuration files in the following directories:
.IP \(bu 3
$XDG_CONFIG_HOME/ladspa_dsp
.IP \(bu 3
$HOME/.config/ladspa_dsp (if $XDG_CONFIG_HOME is not set)
.IP \(bu 3
/etc/ladspa_dsp
.PP
To override the default directories, set the `LADSPA_DSP_CONFIG_PATH'
environment variable to the desired path(s) (colon-separated).
.PP
Each file that is named either \fIconfig\fR or \fIconfig_<name>\fR (where \fI<name>\fR is
any string) is loaded as a separate plugin. The plugin label is either
\fIladspa_dsp\fR (for \fIconfig\fR) or \fIladspa_dsp:<name>\fR (for \fIconfig_<name>\fR).
.PP
Configuration files are a simple key-value format. Leading whitespace is
ignored. The valid keys are:
.TP
.B input_channels
Number of input channels. Default value is 1. May be left unset unless
you want individual control over each channel.
.TP
.B output_channels
Number of output channels. Default value is 1. This parameter is not
currently set automatically because the number of LADSPA ports must be
known before the effects chain is built. Initialization will fail if it
does not match the effects chain.
.TP
.B LC_NUMERIC
Set `LC_NUMERIC' to the given value while building the effects chain.
Default value is `C', which gives consistent number parsing behavior
regardless of the system locale and LADSPA host behavior. Setting this to
an empty value uses the default system locale. The special value `none'
leaves `LC_NUMERIC' up to the LADSPA host (not generally recommended).
.TP
.B effects_chain
String to build the effects chain. The format is the same as an effects
file, but only a single line is interpreted.
.PP
Example configuration:
.EX
	# This is a comment
	input_channels=1
	output_channels=1
	LC_NUMERIC=C
	effects_chain=gain -3 lowshelf 100 1s +3 @/path/to/eq_file
.EE
.PP
Relative file paths in the \fBeffects_chain\fR line are relative to the
directory in which the configuration file resides.
.PP
The loglevel can be set to `VERBOSE', `NORMAL', or `SILENT' through the
`LADSPA_DSP_LOGLEVEL' environment variable.
.PP
Note: The resample effect cannot be used with the LADSPA frontend.
.SS Examples
See https://github.com/bmc0/dsp/blob/master/README.md for usage examples.
.SH BUGS
No support for metadata.
.PP
Some effects do not support plotting.
.PP
When plotting an effects chain containing the \fBnoise\fR effect, a different
random sequence is generated for each output channel regardless of whether the
noise should be correlated between outputs. Summing correlated noise works
correctly.
.SH LICENSE
This software is released under the ISC license.