Find file
Fetching contributors…
Cannot retrieve contributors at this time
478 lines (368 sloc) 21.1 KB
[P]hase [H]armonic [A]dvanced [S]ynthesis [EX]periment
Welcome to the new incarnation of PHASEX!
PHASEX is an experimental MIDI softsynth for Linux/ALSA/JACK with a
synth engine built around flexible phase modulation and flexible
oscillator/LFO sources. Modulations include AM, FM, offset PM, and wave
select. PHASEX comes equipped with multiple filter types and modes, a
stereo crossover delay and chorus with phaser, ADSR envelopes for amplifier
and filter, realtime audio input processing capabilities, and more.
* C Hacker Friendly. PHASEX is written entirely in C. No C++. No
assembly. From the very beginning, the program framework (now
a much larger application framework) has existed to support a synth
engine written in good ol' C. Now the application framework is well
on its way to serving as a basic platform for developing new
instruments written in... you guessed it... C.
* Reliability. Whether playing music on stage, working in the studio,
or writing music on a laptop in a coffee shop, the instruments and
tools need to function properly and consistently in order to achieve
good results. Musicians need to be able to focus on the music, not
shortcomings of the toolset.
* Compatibility, starting with full JACK and ALSA API support (and
eventually support for additional audio / MIDI libraries), allowing
PHASEX to be used in any pro audio workflow on a modern Linux
distribution. The standalone application design was partially
chosen for this reason -- not all sequencers and DAWs support all
instrument plugin formats. For this reason, PHASEX will most likely
see support for features such as ALSA channel mapping, JACK
Freewheel, JACK Session, and even Non Session Manager (NSM) before
seeing support for DSSI or LV2? Why? PHASEX already runs side by
side with Ardour, MusE, Rosegarden, Qtractor, Seq24, or any other
sequencer supporting JACK MIDI or the ALSA sequencer interface, no
matter what plugin APIs these sequencers do or do not happen to
support. In the future, adding support for plugin formats may be
deemed necessary for making PHASEX more useful in certain
situations. For now, the development effort will focus on
fine-tuning and extending the usage of the APIs already supported.
* Stage friendly. Low MIDI latency and low timing jitter. Run with
or without the GUI. Ability to specify alternate config files
(just be sure you have a backup of your ~/.phasex/on-stage.cfg...).
With decent hardware running PHASEX and a recent -rt kernel with
proper permissions, priority tuning, hardware tuning (BIOS settings,
PCI latencies, etc.), and audio buffer tuning, MIDI-audio latencies
of under 10ms should be well within reach for audio hardware capable
of running with 128 sample buffer periods or better.
* Studio friendly. Focus on workflow. The recent development work
has been aimed at making PHASEX easier to use and efficient to
manage (quick configuration, built in session management, MIDI
connections in the menus, etc.). Again, the focus should be on the
music, not on remembering how to make the instrument jump through a
bunch of hoops just to do it's job. PHASEX development will
continue in this direction with new session management features
starting with storing audio and MIDI connections, midimaps, and
configs with session saves, and optionally recalling these with
session loads.
* Great sound. This one is of course, quite subjective, but it still
comes down to the same juggling act on the programming end: Choosing
the "right" algorithms, making resource usage vs. sound quality
decisions, calculation speed vs. accuracy decisions, and of course,
decisions about how much programming time is spent on what features
(hmmm... is it worth the time to do the math and write and debug the
code for nifty-feature-x, or is time better spent fixing
broken-feature-y?). At this point, a lot of these decisions have
been made and the synth engine as a whole is very stable, but these
things are not necessarily set in stone. There is always room for
improvement by researching new ways of doing the math and learning
new DSP tricks for solving some of the flaws inherent in the current
methods. So, how does it sound? Decide for yourself.
* GTK Interface:
PHASEX has a GUI, complete with knobs and buttons for all 150+
patch parameters and switchable color themes. True fullscreen
mode, notebook, one-page, and widescreen layouts, and select-able
fonts and knob sizes allow PHASEX to fit into any amount of screen
real-estate (now able to squeeze into sizes smaller than 800x600,
or more comfortably sit in a 1920x1080 desktop).
* Polyphonic:
PHASEX is polyphonic. In poly mode, voices are allocated
dynamically up to the runtime configurable voice limit.
Oscillators and filters function independently per-voice. LFOs,
however, function on a per-part basis. In addition to Poly mode,
the three legacy Mono modes are still supported: Mono Smooth mode
puts new notes back into sustain if playing legato. Mono Multikey
mode maps the keys played, in order, round-robin style to all
oscillators based on MIDI note frequencies in use. In Mono
Retrigger mode, envelopes and oscillator init phases are
retriggered every time a note is played.
* Multitimbral:
Many parts of PHASEX were re-architected during the v0.14.x
development phase. This release is now fully multitimbral, with
each part running in its own thread. The number of parts (and
thus engine threads) is determined at compile time, and has been
tested rigorously with 8 voices on a quad-core q6600. Part
switching in the GUI is clean and efficient. Session handling
support takes the fuss out of keeping track of session patches.
* ALSA Audio:
PHASEX supports native ALSA for flawless xrun free audio playback.
Sample rate and buffer size are easily configurable to achieve the
lowest possible latencies for your system.
* JACK Audio:
PHASEX has supported JACK audio from the beginning of the public
releases. Any sample rate or buffer size used by jackd can be
utilized. Both single-stereo-pair and multi-stereo-pair routings
are supported. Basic Transport is supported, but as of this time,
the tempo and LFO sync are largely untested and still need some
The ALSA sequencer interface has been with PHASEX since its first
public release. As of v0.14.x, the ALSA raw MIDI interface is
also supported, offering lower timing jitter when controlling
PHASEX via hardware MIDI.
JACK MIDI provides 100% sample-accurate timing by design. PHASEX
has embraced this design, and the new underlying MIDI system is
the result.
* Generic MIDI:
Generic MIDI (/dev/midi) support is provided for communicating
with MIDI devices or drivers that use a standard POSIX interface
supporting open/close/poll/read/fcntl on a named pipe, socket, or
device file. The generic MIDI support has only been tested with
ALSA's /dev/midi* and /dev/snd/midiC?D? device files (which are
accessed through the POSIX interface, not the alsa-lib API), but
should perform the same with any MIDI device interface as long as
additional system and library calls are not needed. Currently,
compatibility with serial to MIDI converters is unknown, so please
send feedback (positive or negative) if you attempt to use a
serial to MIDI converter.
* MIDI Control:
All parameters are quickly mappable and ready for full realtime
control. Parameter names are right clickable for quick and easy
MIDI controller assignments. Each MIDI controller can be mapped
to up to 15 parameters. Velocity sensitivity and key aftertouch
can modulate the amp and the filter, or used as raw oscillators of
LFOs. MIDI maps may be saved and later restored to facilitate
switching between different working environments.
* Realtime MIDI Clock:
Addresses the issue of how to accurately synchronize MIDI events
with individual frame positions within a buffer period instead of
quantizing the timing to whatever the audio buffer period happens
to be. Typical audio hardware supports stable realtime operation
at anywhere from 2-20 milliseconds latency. With no additional
synchronization, this audio latency becomes a self-imposed
quantization factor on event timing, adding MIDI timing jitter
roughly equal to the duration of one audio buffer processing
period, hence the need for a MIDI clock:
A timerless, tickless audio / MIDI timing synchronization system.
The clock latches onto the audio buffer processing periods and
generates a high-precision MIDI clock reference, with much less
jitter than timing based on when the audio processing cycle
actually wakes up (which can really be anywhere within the
period). This clock is agnostic to the specific audio and MIDI
drivers being used and requires no timer interrupts. In essence,
the timing accuracy achieved is only dependent on the MIDI
hardware/software and realtime scheduling latencies of the system.
With decent professional MIDI and audio hardware, sub-millisecond
timing accuracy is available with total MIDI event latency under
10ms, accurate and responsive enough for most professional sound
sound engineers and musicians.
* Multithreaded:
PHASEX uses separate threads with realtime scheduling for audio
output, MIDI input, and the synth engine. The GTK UI runs in its
own thread at normal scheduling priority.
* 4 Oscillators:
Each oscillator features wave shape selection (including sampled
and generated waveforms), bipolar or unipolar output, frequency
source selection (midi note, tempo based, tempo based w/ phase
trigger, audio input, or velocity/aftertouch), mix modulation
supporting standard or AM mixing, per-osc transpose and
pitchbender amounts, and select-able modulation sources for AM, FM,
phase offset, and waveshape modulation. For all but waveshape
modulation, both oscillators and LFOs are available as modulation
* 4 LFOs:
The dedicated per-osc LFOs have been replaced with generic
per-part LFOs. The LFOs are identical to the oscillators with the
following exceptions: They are per-part instead of per-voice,
they do not accept other modulation sources (yet), they are not
transposed by the master transpose parameter, and when sourced to
velocity, aftertouch data is ignored. New routing capabilities
have been added with the Voice AM, LFO-1 Rate, Cutoff, and LFO-3
Rate per-LFO dedicated function parameters.
* ADSR Envelopes:
PHASEX comes equipped with standard ADSR envelopes for the
amplifier and filter. Duration times follow a parabolic curve.
The amplitude of the amplifier envelope follows a logarithmic
curve. The filter envelope is linear with respect to the musical
scale. Both envelopes can also be used as oscillator and LFO
* Filter:
The first filter style is a Chamberlin filter (2-pole 12db/octave)
hacked up so it doesn't destabilize at high cutoff or resonance
values. Low pass, high pass, band pass, and band stop modes are
available, as well as four combo modes. The filter has it's own
ADSR envelope (identical to amplifier envelope with the added
option of switching to negative polarity). One of the LFOs may be
mapped to modulate filter cutoff. Cutoff controller values map
directly to MIDI note numbers. Optional keyfollow mode allows
cutoff to follow the key played for the MIDI note, or the lowest,
highest, or most recent key in play. The filter is per-voice in
Poly mode.
As of v0.14.x, a 4-pole 24db/octave Stilson/Smith style Moog
filter has been added. In PHASEX style, both a distorted and a
clean version are available.
The 3-pole filter is experimental, and might not be included in
future versions. Future plans include fixing the frequency
calculation and turning this into a proper 303 style filter.
* Chorus:
The PHASEX stereo crossover chorus (inspired by the chorus on the
old Akai S-series samplers) adds depth and that extra bit of
stereo separation for an even wider stereo field.
* Delay:
The delay sports the normal mix and feedback controls, along with
a channel crossover option. A dedicated LFO allows modulation of
playback position, great for flange-like effects (or really
messing with people's heads at longer delay times). Delay times
are select-able based on tempo, in both 1/64th and 1/48th
* Input Processing:
Up to two JACK input channels (as mono, dual mono, or stereo) may
be used as audio sources for both oscillators and LFOs. An input
envelope follower and input boost allow for sources like guitar or
vocals to be processed with ease.
* Stereo Width:
The width of the stereo image is adjustable via a MIDI controller
from pure mono to pure stereo. Can be used during sound design to
help create patches that sound very different when a center
playback channel is present. Very useful for getting patches to
sit in the mix properly.
* Velocity Sensitivity:
Velocity of Note-On events can modify the filter or amplifier, and
can also be routed through the LFOs.
* ALSA >= 1.0.18
* JACK (JACK or JACK2/jackdmp should work).
* libsamplerate
* GTK-2 (version 2.8 or later recommended).
* Realtime Linux Kernel (latest 3.0.x-rt or newer recommended).
* 1.6 GHz or faster CPU (multiple core, 64-bit preferred).
* LASH >= 0.5.4.
* libuuid (required by LASH).
For most installations, all that should be needed is the following:
cd phasex-0.14.94
./configure --enable-arch=native
make install
Other useful configure flags are --enable-debug=, --enable-32bit,
--enable-cpu-power=, and enable-parts=.
See INSTALL for full compilation and installation instructions.
Usage: phasex [options] [<patch>[,<patch>[,<patch>[,...]]]]
-c, --config-file= Alternate config (takes precedence over default
config file and command line options.
-A, --audio-driver= Audio driver: alsa or jack.
-a, --audio-device= ALSA device name (hw:x,y format).
-r, --sample-rate= Audio sample rate (ALSA only).
-p, --latency-periods= Number of buffer latency periods (ALSA only).
-s, --period-size= Buffer period size (power of 2, ALSA only).
-M, --midi-driver= MIDI driver: jack, alsa-seq, alsa-raw, generic.
-m, --midi-port= MIDI input port or device name (driver specific).
-f, --fullscreen Start GUI in fullscreen mode.
-x, --maximize Start GUI with main window maximized.
-b, --bpm= Override BPM in patch bank and lock BPM parameter.
-t, --tuning= Base tuning frequency for A4 (default 440 Hz).
-i, --input= Comma separated pair of audio input matches.
-o, --output= Comma separated pair of audio output matches.
-O, --oversample Use double the sample rate for internal math.
-U, --undersample Use half the sample rate for internal math.
-G, --no-gui Run PHASEX without starting the GUI.
-d, --debug= Debug class (Can be repeated. See debug.c).
-l, --list Scan and list audio and MIDI devices.
-v, --version Display version and exit.
-h, --help Display this help message and probe ALSA hardware.
LASH Options:
-P, --lash-project= LASH project name.
-S, --lash-server= LASH server address.
-I, --lash-id= LASH client ID.
-L, --disable-lash Disable LASH completely for the current session.
Please note that unless an alternate config file is used, all command
line options take precedence over settings saved in the config file.
If using phasex with JACK, make sure jackd is running (and configured
for your system) before running phasex. If jackd is not running,
PHASEX will fall back to using the ALSA drivers. If you use software
like qjackctl to handle your audio and MIDI patchbay, then the command
line port connection options aren't necessary. The -i and -o options
accept comma separated substring pairs to match on JACK audio port
names. The -m option takes a comma separated list of MIDI ports (X:Y
for ALSA Seq, hw:X,Y for ALSA Raw, or device file for generic raw
MIDI). For example:
phasex -A jack -i _1,_2 -o _1,_2 -M alsa-seq -m 16:0,20:0
will match JACK ports alsa_pcm:capture_1 and alsa_pcm:capture_2 for
input, alsa_pcm:playback_1 and alsa_pcm:playback_2 for output, and the
first two ALSA sequencer hardware MIDI input ports with relatively
recent versions of JACK and ALSA. (Jack2/jackdmp uses 'system'
instead of 'alsa_pcm'. ALSA port designations vary from system to
To view a full technical description of any parameter, simply
middle-click on the parameter label. Similarly, a right-click on any
parameter label allows on-the-fly mapping of any MIDI controller to
any parameter (up to 15 parameters may be mapped to a single MIDI
controller). All remaining features are directly visible and
accessible through the GUI and its (very short) menus.
Patch bank storage works like it does on just about any modern
hardware synth, with one exception: Patches themselves are stored by
name in the filesystem, which are then mapped by program number in the
patch bank. The individual program slots in the patch bank simply
reference the stored patches by name. In Autosave mode, a patch is
automatically saved before a new patch is loaded during a program
change without any user intervention. Autosave mode is intended for
use during sound design sessions (and should be avoided the rest of
the time to keep from overwriting patches). In Warn mode, warnings
will be issued will be issued when unsaved modifications to the patch
will be lost. In Protect mode, program changes always take effect
immediately, so any patch modifications are quietly discarded and
patches never get (re)saved unless they are saved explicitly.
In the event of catastrophic failure, and PHASEX fails to start, edit
~/.phasex/phasex.cfg and look for the last settings change, or simply
remove phasex.cfg (but not any directories) and PHASEX will start with
the default config and auto-detect your setup all over again.
PHASEX is distributed under the terms of the GNU Public License,
version 3. See LICENSE for details.
Copyright (C) 1999-2012 William Weston <>.
With portions of and contributions to the PHASEX source:
Copyright (C) 2010 Anton Kormakov.
Copyright (C) 2007 Peter Shorthose.
Copyright (C) 2004 Sean Bolton.
Copyright (C) 1999 Tony Garnock-Jones.
See AUTHORS for details.
A new PHASEX web site and git repository are coming soon. Until then,
please download PHASEX using the links provided in the mailing lists.
The primary goal of the PHASEX project is to offer a free software
synthesizer for GNU/Linux/ALSA/JACK systems that is useful in a MIDI
production environment. Please send any kind of feedback you will for
What do you like about PHASEX?
What don't you like about PHASEX?
What features do you think are missing?
What would you do differently?
What are you looking for in a UI?
Is there anything preventing you from using PHASEX dependably on stage?
Is there anything preventing you from using PHASEX dependably in the studio?
Anything you say can only influence me to make PHASEX better.
Kind Regards,
--William Weston <>