MAN page from StartCom 5 timidity++-2.13.2-1.2.1.i386.rpm


Section: User Commands (1)
Updated: April 25 2004


TiMidity++ - MIDI-to-WAVE converter and player 


timidity[options] filename [...] 


TiMidity++ is a converter that converts some of MIDI files(supported formats: Standard MIDI files (*.mid), Recomposer files(*.rcp, *.r36, *.g18, *.g36) and Module files (*.mod)) into formattedaudio files (e.g. RIFF WAVE). TiMidity++ uses GravisUltrasound-compatible patch files or Soundfonts (*.sfx, *.sf2) togenerate digital audio data from MIDI files. The digital audio datagenerated by TiMidity++ can be stored in a file for processing,or played in real time through an audio device.
In real time playing, TiMidity++ can show the lyrics containedin KAR or WRD files. 


You can use the following expressions as the filename argument:
Read a MIDI file from standard input.
Read a MIDI file from the specified path on a filesystem.

Read and play all MIDI files in the specified directory. Forexample,

% timidity some/where/

plays all files in the directory some/where/.

Archive File
Extract and play the file(s) in the archive. If you want to specify acertain MIDI file in the archive, append #<MIDI-filename> tothe archive name. The path after `#' allows the use of the wildcardexpressions (case insensitive).
You can use escape sequence \xHH, where `HH' is a ASCII numberin hexadecimal integer.

For example:

% timidity
Plays file.mid in
% timidity file.lzh#*.mid
Plays any files that match the wildcard expression *.mid infile.lzh
% timidity file.tgz#*
This expression is the same as file.tgz

Since these mechanism are contained in TiMidity++ itself, youcan use this syntax even in the MS Windows environment.

TiMidity++ can handle the following archive formats:

tar (*.tar)
tar+gzip (*.tar.gz, *.tgz)
zip (*.zip)
lzh (*.lzh, *.lha)
(lh0, lh1, lh2, lh3, lh4, lh5, lh6, lz4, lzs and lz5 are available)

Other archives can be expanded if expander command is specified on thecompile phase. TiMidity++ will pipe that command.


Play the MIDI file in the specified article on the news server. If anewsgroup is specified TiMidity++ plays all MIDI files found inany article posted to that newsgroup.
TiMidity++ parses MIME Multi-part messages in case ofnews://* scheme, extracts MIDI file that was post to that group,or in case of normal file, you can explicitly handle MIME documents bynaming that file with *.mime extension, or prefix that file withmime:

The following MIME-types are allowed:

uu-encoded file
is required
base64 encoded
Content-Transfer-Encoding: base64
is required
Content-Transfer-Encoding: quoted-string
is required
Mac BinHex format
only HQX format is available

Play the file specified in the URL.

For example:

% timidity
plays the specified MIDI-file directly from the network.

If these expression are used in the *.cfg files, you can even usepatch-files (and others) from remote machines.



TiMidity++ can handle the following file formats:
.mid, .rmi (Format 0, 1, 2)
Standard MIDI File
.rcp, .r36, .g18, .g36 (Recomposer formats)
Recomposer format which is product for COME ON MUSIC co.
.mfi (MFi Version 3 - Melody Format for i-Mode)
i-Mode is Japanese local mobile phone
.kar (Karaoke format)
Displays the lyrics as a Lyric Meta Event message.
.mod, mod.* (Module file)
.wrd (WRD format)


The following command line options are accepted by TiMidity++:
-A [n][,m](a)



Multiplies the master volume by n%. Default value is 70%.Higher amplification makes louder sounds. You can specify the drumpower, ratio of drum volume from the other channels. The allowedvalues of amplification range from 0 (no sound) to 800.
Optionally to put `a' character along with -a option, or to use--volume-compensation, instructs TiMidity++ toregularize the volume. You can easily gain dynamic range.

For example:

volume 90%, drum power 100%, compensation is off
volume 70%, drum power 120%, compensation is off
volume 90%, drum power 120%, compensation is off
volume 70%, drum power 100%, compensation is on
volume 90%, drum power 100%, compensation is on
volume 70%, drum power 120%, compensation is on
volume 90%, drum power 120%, compensation is on
-a, --[no-]anti-alias
Turns on antialiasing. Samples are run through a lowpass filterbefore playing, which reduces aliasing noise at low resamplingfrequencies.
-B n,m, --buffer-fragments=n,m
For the Linux/FreeBSD/OSS/ALSA/Windows sound driver, selects thenumber of buffer fragments in interactive mode. Increasing the numberof fragments may reduce choppiness when many processes are running.It will make TiMidity++ seem to respond sluggishly to fastforward, rewind, and volume controls, and it will throw the statusdisplay off sync. Specify a fragments number of 0 to use the maximumnumber of fragments available.
-C n, --control-ratio=n
Sets the ratio of sampling and control frequencies. This determineshow often envelopes are recalculated -- small ratios yield betterquality but use more CPU time.
-c file, --config-file=file
Reads an extra configuration file.
-D n, --drum-channel=n
Marks channel as a drum channel. If channel is negative, channel-n is marked as an instrumental channel. If n is0, all channels are marked as instrumental.
-d dir, --interface-path=dir
Specifies the directory containing installed dynamic-link interfacemodules.
-E mode, --ext=mode
Set TiMidity++ extend modes. The following modes are available(capitalized switch means disable this feature):
w/W, --[no-]mod-wheel
Enable/disable modulation controlling.
p/P, --[no-]portamento
Enable/disable portamento controlling.
v/V, --[no-]vibrato
Enable/disable NRPM vibration.
s/S, --[no-]ch-pressure
Enable/disable channel pressure controlling.
e/E, --[no-]mod-envelope
Enable/disable modulation envelope controlling.
t/T, --[no-]trace-text-meta
Enable/disable tracing all Text Meta Events.
o/O, --[no-]overlap-voice
Accept/reject pronouncing multiple same notes.
z/Z, --[no-]temper-control
Enable/disable MIDI Tuning Standard temperament controlling.
mHH, --default-mid=HH
Sets the manufacturer ID to HH (where HH are twohex-digits).
HH values of GS/gs, XG/xg or GM/gm areunderstood as 41, 43 and 7e respectively.
MHH, --system-mid=HH
Sets the system manufacturer ID to HH (where HH are twohex-digits).
In this option, the manufacuture ID is set unchangable. ManufacutureID from the input file would be ignored.
bn, --default-bank=n
Use tone bank n as the default.
Bn, --force-bank=n
Sets the bank number of all channels to n.
in[/m], --default-program=n[/m]
Use the program number as the default instrument. Any Program Changeevents in MIDI files will override this option.
If n is followed by /m the default program number of thechannel m is specified by n.
In[/m], --force-program=n[/m]
Similar to -Ei but this ignores all program changes.
F args
For effects. See below. In args option, you can specifyfollowing effect options:
delay=(d|l|r|b)[,msec], --delay=(d|l|r|b)[,msec]
Sets delay type.
d, 0
Disabled delay effect.
l, 1
Left delay.
r, 2
Right delay.
b, 3
Swap left & right.

Optional msec is the delay time.

chorus=(d|n|s)[,level], --chorus=(d|n|s)[,level]
d, 0
Disable this effect.
n, 1
Enable MIDI chorus effect control.
s, 2
Surround sound, chorus detuned to a lesser degree (default).

The optional parameter level specifies the chorus level 0to 127.

reverb=(d|n|g|f|G)[,level], --reverb=(d|n|g|f|G)[,level]
d, 0
Disable MIDI reverb effect control.
n, 1
Enable MIDI reverb effect control. This effect is only available instereo.
g, 2
Global reverb effect.
f, 3
Freeverb MIDI reverb effect control (default).
G, 4
Global freeverb effect.

The optional parameter level specifies the reverb level 0to 127.

vlpf=(d|c|m), --voice-lpf=(d|c|m)
d, 0
Disable LPF effect.
c, 1
Chamberlin resonant LPF (12dB/oct) (default).
m, 2
Moog resonant low-pass VCF (24dB/oct)
ns=n, --noise-shaping=n
Enable the n th degree noiseshaping filter. The distortion atdecay stage is improved, but the noise on human auditory feelingincreases because it shifts to a high frequency. In case of 8-bitlinear encoding, valid values of n are in the interval from0 (min) to 4 (max). Default value is 4. In case of16-bit linear encoding, valid values of n are in the interval from0 to 4. According to the value, it works as following.Default value is 4.
No noise shaping.
Traditional noise shaping.
Overdrive-like soft-clipping + new noise shaping.
Tube-amplifier-like soft-clipping + new noise shaping.
New noise shaping.
resamp=(d|l|c|L|n|g), --resample=(d|l|c|L|n|g)
d, 0
No interpolation.
l, 1
Linear interpolation.
c, 2
Cubic spline interpolation.
L, 3
Lagrange method.
n, 4
Newton polynomial interpolation.
g, 5
Modified Gauss effect (default).

This option affects the behavior of -N option.

-e, --evil
Make TiMidity++ evil. For the Win32 version, this increases thetask priority by one. It can give better playback when you switchtasks at the expense of slowing all other tasks down.
-F, --[no-]fast-panning
Turns on fast panning to accommodate MIDI pieces that expect panningadjustments to affect notes that are already playing. Some files thatdon't expect this have a habit of flipping balance rapidly betweenleft and right, which can cause severe popping when the -F flagis used.
In the current version of TiMidity++ this option is a toggle.
-f, --[no-]fast-decay
Toggles fast envelopes. This option makes TiMidity++ faster butthe release time of the notes are shortened.
-g sec, --spectrogram=sec
Open the Sound-Spectrogram window. This option is activated if thesystem has support for the X Window System.
-H n, --force-keysig=n
Specify the number of key signature. MIDI file is performed withtransposition to the key of the number of sharp (when n ispositive) or flat (when n is negative). Valid values of nare in the interval from -7 to 7. For example, if nis 1, MIDI file with any tonality will be performed transposingto 1 flat (i.e., F major or D minor). Therefore, it isconvenient to use for the accompaniment when practicing a musicalinstrument by the tonality which is easy to perform.
-h, --help
Show the help message.
-i mode, --interface=mode


Selects the user interfaces from the compiled-in alternatives.mode must be begun with one of the supported interfaceidentifiers. Run TiMidity++ with the -h option to see alist.
For ALSA sequencer interface, optionally to use--realtime-priority, set the realtime priority by n,and to use --sequencer-ports, set the number of openedsequencer ports. Default value is 4.
The following identifiers may be available:
dumb interface
ncurses interface
S-Lang interface
X Athena Widget interface
Tcl/Tk interface
Motif interface
vt100 interface
Emacs interface
(use ``M-x timidity'' in Emacs)
skin interface
Environment variable TIMIDITY_SKIN must be set to the path ofthe skin data (compressed data are also supported).
GTK+ interface
Launch TiMidity++ as MIDI server.
Launch TiMidity++ as ALSA sequencer client.
Windodws synthesizer interface
Windodws GUI interface
PortMIDI synthesizer interface
UMP interface
Interface options
Option characters may be added immediately after the interfaceidentifier. The following options are recognized:
v, --verbose=n
Increases verbosity level. This option is cumulative.
q, --quiet=n
Decreases verbosity level. This option is cumulative.
t, --[no-]trace
Toggles trace mode. In trace mode, TiMidity++ attempts todisplay its current state in real time. For the Linux sound driver,this is accomplished through the use of short DMA buffer fragments,which can be tuned via the -B option.
l, --[no-]loop
Loop playing (some interfaces ignore this option)
r, --[no-]random
Randomize file list arguments before playing
s, --[no-]sort
Sort file list arguments before playing
D, --[no-]background
Daemonize TiMidity++ in background (for alsaseq only)
-j, --[no-]realtime-load
Enable the loading of patch files during play.
-K n, --adjust-key=n
Adjusts key (i.e., transposes the song) by n half tones. Rangesfrom -24 to 24.
-k msec, --voice-queue=msec
Specify audio queue time limit to reduce voices. If the remainingaudio buffer is less than msec milliseconds, TiMidity++tries to kill some voices. This feature makes it possible to playcomplicated MIDI files on slow CPUs. Setting msec to zero tellsTiMidity++ to never remove any voices.
-L path, --patch-path=path
Adds path to the library path. Patch, configuration, and MIDIfiles are searched along this path. Directories added last will besearched first. Note that the current directory is always searchedfirst before the library path.
-M name, --pcm-file=name
TiMidity++ can play a PCM file instead of a MIDI file. If``auto'' is specified, TiMidity++ tries to open foo.mid.wav orfoo.mid.aiff when playing foo.mid. If ``none'' is specified, thisfeature is disabled. Otherwise just plays name.
-m msec, --decay-time=msec
Modify envelope volume decay time. msec is the minimum numberof milliseconds to sustain a sustained note.
Disable sustain ramping, causes constant volume sustains (default).
Effectively behaves as if all sustains are ignored, volume ramping isthe same as normal stage 3.
A note at full volume will decay for 3 seconds once it begins tobe sustained (assuming the regular stage 3 rate would not cause it todecay even longer). Softer notes will of course die sooner.
-N n, --interpolation=n
Sets interpolation parameter. This option depends on the-EFresamp option's value.
cspline, lagrange
Toggles 4-point linear interpolation (default is on).
n point interpolation using Newton polynomials. n must bean odd number from 1 to 57.
n+1 point modified Gauss interpolation. Ranges 0(disable) to 34 (max), default to 25.

In either way, linear interpolation is used if audio queue < 99%.

-O mode, --output-mode=mode











Selects the output mode from the compiled-in alternatives.mode must begin with one of the supported output modeidentifiers. Run TiMidity++ with the -h option to seethe list.
Special in Ogg FLAC output mode, verifying generated data (will be abit slower), the size of header padding (default is 4096), thecompression level (0 to 8) (default is 5), and enabling OggFLAC streamcan be specified by --flac-verify, --flac-padding,--flac-complevel and --oggflac options respectively.
Special in Ogg Speex output mode, the compression quality (0 to 10)(default is 8), Enabling VBR output, enabling ABR output and settingthe ratio to n, enabling VAD (voice activity detection), enabling DTX(discontinuous transmission), the encoding complexity (0 to 10)(default is 3), and frames in a single Ogg packet (0 to 10) (defaultis 1) can be specified by --speex-quality,--speex-vbr, --speex-abr, --speex-vad,--speex-dtx, --speex-complexity and--speex-nframes options respectively.
The following identifiers are available in all versions:
Outputs via audio device (default)
Output to ALSA
Generate raw waveform data. All format options are supported. Commonformats include:
16-bit signed linear PCM
8-bit unsigned linear PCM
Generate Sun Audio (au) data
Generate AIFF data
Generate RIFF WAVE format output. If output is directed to anon-seekable file, or if TiMidity++ is interrupted beforeclosing the file, the file header will contain 0xffffffff in the RIFFand data block length fields. The popular sound conversion utilitysox is able to read such malformed files, so you can pipe datadirectly to sox for on-the-fly conversion to other formats.
List MIDI events
MOD -> MIDI conversion
Ogg Vorbis
Ogg Speex
Format options
Option characters may be added immediately after the mode identifierto change the output format. The following options are recognized:
S, --output-stereo
M, --output-mono
s, --output-signed
Signed output
u, --output-unsigned
Unsigned output
1, --output-16bit
16-bit sample width
2, --output-24bit
24-bit sample width
8, --output-8bit
8-bit sample width
l, --output-linear
Linear encoding
U, --output-ulaw
u-Law (8-bit) encoding
A, --output-alaw
A-Law encoding
x, --[no-]output-swab
Byte-swapped output

Note that some options have no effect on some modes. For example, youcannot generate a byte-swapped RIFF WAVE file, or force uLaw outputon a Linux PCM device.

-o file, --output-file=file
Place output on file, which may be a file, device, or HP-UXaudio server, depending on the output mode selected with the -Ooption. The special filename `-' causes output to be placed onstdout.
-P file, --patch-file=file
Use patch file for all programs.
-p [n](a)


Sets polyphony (maximum number of simultaneous voices) to n.
Optionally to put `a' character along with -p option, or to use--polyphony-reduction, instructs TiMidity++ to enableautomatic polyphony reduction algorithm.
-Q n[,...](t)


Cause channel n to be quiet. n can carry out packagespecification by `,'. If n is 0, all channels are turnedoff. Continuously, specifying -n, channel n is turnedback on.
On the other hand, to put `t' character after -Q option or touse --temper-mute describes temperament mute. This muteschannels of specific temperament type n. For presettemperament, n can range 0 to 3. For user-definedtemperament, n can range 4 to 7.
-q sec/n, --audio-buffer=sec/n
Specify audio buffer in seconds. sec maximum size of buffer,n percentage filled at the beginning (default is 5.0/100)(size of 100% equals the whole device buffer size).
-R msec
Enables Pseudo Reverb Mode. It sets every instrument's release tomsec ms. If msec is 0, msec is set to800 (default).
-S n, --cache-size=n
Sets the re-sample cache size to n bytes. If n equals0 any sample caches are disabled. The default value ofn is 2097152 (2MB).
-s freq, --sampling-freq=freq
Sets the resampling frequency (Hz or kHz). Not all sound devices arecapable of all frequencies -- an approximate frequency may beselected, depending on the implementation.
-T n, --adjust-tempo=n
Adjust tempo to n%; 120 play MOD files with an NTSCAmiga's timing.
-t code, --output-charset=code
Sets output coding of Japanese text. Possible values of codeare:
determined by the LANG environment variable.
Translates non-ASCII code to period.
No conversion.
Convert from windows-1251 to koi8-r.
Outputs EUC (Japan) coding.
Outputs JIS coding.
Outputs SJIS coding.
-U, --[no-]unload-instruments
Unload all instruments from memory between MIDI files. This canreduce memory requirements when playing many files in succession.
-V power, --volume-curve=power
Set the power of volume curve. The total amplification becomesvolume^power. 0 (default) uses the regular tables. Anynon-zero value causes all midi to use the new user definedvelocity/volume/expression curve (linear: 1, ideal:~1.661, GS: ~2).
-v, --version
Show the version string
-W mode, --wrd=mode
Play WRD file.

Allowed values of mode are:

X Window System mode
Windows console mode
TTY mode
Dumb mode (outputs WRD events directory)
do not trace WRD
Sets WRD options:
Sets the WRD options. an is the name of option and bn isthe value.
Emulates timing (@WAIT, @WMODE) bugs of the original MIMPI player.The emulation levels are:
do not emulate any bugs of MIMPI
only emulate some bugs (default)
emulate all known bugs
Use file as WRD file only no file matching *.wrd is found.
Uses file as WRD file.

WRD mode must also use trace mode (option -i?t) orthe timing of the WRD events will be terrible.

-w mode, --rcpcv-dll=mode
Extended mode for MS Windows. The following options are available:
-w r
Use rcpcv.dll to play RCP/R36 files.
-w R
Do not use rcpcv.dll (default).
-x str, --config-string=str
Configure TiMidity++ with str. The format of str isthe same as timidity.cfg.

For example:
-x'bank 0\n0 violin.pat'
Sets the instrument number 0 to violin.
Character `\' (Ascii 0x5c) in the str is treated as escapecharacter like in C literals. For example \n is treated ascarriage return.

-Z file, --freq-table=file
Cause the table of frequencies to be read from file. This isuseful to define a tuning different from 12-equal temperament. If``pure'' is specified, TiMidity++ plays in trial pure intonation.
-Zpure[n(m)], --pure-intonation=[n(m)]
Play in trial pure intonation by Key Signature meta-event in the MIDIfile. You can specify the initial keysig by hand, in case the MIDIfile does not contains the meta-event. Optionally, n is thenumber of key signature. In case of sharp, n is positive. Incase of flat, n is negative. Valid values of n are in theinterval from -7 to 7. In case of minor mode, you shouldput `m' character along with -Zpure option.
Simulate behavior of specific synthesizer module as much as possible.For the moment, the value of n defined is as follows:
TiMidity++ Default
Roland SC-55
Roland SC-88
Roland SC-88Pro
Roland SC-8850
Reserved for GS family
Reserved for XG family
SoundBlaster Live!
SoundBlaster Audigy
Reserved for other synthesizer modules
TiMidity++ Special 1
Reserved for TiMidity++ specification purposes
TiMidity++ Debug


lsmidiprog(1), mididump(1), patinfo(1), sf2text(1), wav2pat(1),timidity.cfg(5) 


Copyright (C) 1999-2004 Masanao Izumo <>
Copyright (C) 1995 Tuukka Toivonen <>

The original version was developed by Tuukka Toivonen <>until the release of TiMidity-0.2i. His development was discontinuedbecause of his being busy with work.

This program is free software; you can redistribute it and/or modifyit under the terms of the GNU General Public License aspublished by the Free Software Foundation; either version 2 of theLicense, or (at your option) any later version.

This program is distributed in the hope that it will be useful, butWITHOUT ANY WARRANTY; without even the implied warranty ofMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNUGeneral Public License for more details.

You should have received a copy of the GNU General Public Licensealong with this program; if not, write to the Free SoftwareFoundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307USA 


The latest release is available on the TiMidity++ Page,



Eats more CPU time than a small CPU-time-eating animal.

This man page was translated from Japanese to English by me withpoor English skill :-) 


Version 0.2i and earlier:
Tuukka Toivonen <>
Vincent Pagel <>
Takashi Iwai <>
Davide Moretti <>
Chi Ming HUNG <>
Riccardo Facchetti <>
IZUMO Masanao <>
HARADA Tomokazu <>
YAMATE Keiichirou <>
KIRYU Masaki <>
AOKI Daisuke <>
KOYANAGI Masaaki <>
IMAI Kunihiko <>
NOGAMI Takaya <>
WATANABE Takanori <>
TAKEKAWA Hiroshi <>
NAGANO Daisuke <>
KINOSHITA kosuke <>
SHIGEMURA Norikatsu <>
ARAI Yoshishige <>
Glenn Trigg <>
Tim Allen <>
Michael Haardt <>
Eric A. Welsh <>
Paolo Bonzini <>
KIMOTO Masahiko <>
IWAI Takashi <>
Saito <>
SATO Kentaro <>
TAMUKI Shoichi <>
URABE Shohei <>
SUENAGA Keishi <>
SUZUKI Koji <>

(titles omitted and an order different)

and other many people sends information and bug-fix codes.

The English version of this man page was written by NAGANO Daisuke<>.

Now, TAMUKI Shoichi <> and URABE Shohei<> are maintaining the man page.
If you have any comments or suggestions or complaints :) about thisman page, please tell us it.




This document was created byman2html,using the manual pages.