home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Cubase Magazine 26
/
Issue #26.iso
/
2-SOFTWARE
/
AUDIOTOMIDI 1.06
/
README.ENG.TXT
< prev
next >
Wrap
Text File
|
1999-12-14
|
18KB
|
312 lines
REAL TIME SOUND TO NOTE CONVERTER
- AudioToMidi -
(Freeware ver.1.01)
1.0 FEATURES
2.0 MINIMUM SYSTEM REQUIREMENTS
3.0 INSTALLATION
4.0 OPERATION PRINCIPLE
5.0 OPERATION
6.0 LICENSE
7.0 CONTACT
1.0 FEATURES
The present software allows the conversion of a standard audio
signal to MIDI signal with an insignificant delay. The resulting MIDI signal
can be given to a standard MIDI device, PC speaker, built-in sequencer. The
audio signal spectrum is displayed in a special window in real time. The
present software also allows special options:
-Input audio signal volume normalization;
-Input audio signal correction by built-in graphic equalizer;
-Consideration of possible audio signal frequency deviation from the
standard note frequencies, e.g., because of the difference between the guitar
sound and the tuning fork;
-Note sensor selectivity meaning the sensor sensibility of a particular
note to the adjacent note signals. The graphic representation of the
selectivity is provided. The parameter of this option impacts the conversion
delay time.
-Note volume filter allowing to ignore low notes and noise;
-Output note filter, allowing to ignore the notes of pre-set loftiness
when MIDI signal generation. The note interval and/or presumed key can
also be set.
-Note duration filter allowing to ignore accidental short notes when
writing to the built-in sequencer.
-Graphic simulation of the first four harmonics of recognized
instrument or voice;
-MIDI signal transposition. Resulting note loftiness shift at the
integer number of semi-tones.
-MIDI instrument selection when signal generation;
-Resulting note volume selection.
-MIDI channel selection when MIDI signal generation.
-Monophonic mode which allows the lowest note selection from a
number of simultaneous notes, thus providing the separation of the first
instrument/voice harmonic, and errorless conversion of monophonic
melodies.
The general setting is automatically saved. Separate saving, opening
and resetting of the equalizer, harmonic model and filter settings are
provided.
The conversion result is presented in real time by highlighted piano
keys. The piano keyboard window can also be used to generate the sound
corresponding to the pressed key.
The signal spectrum representation allows the software usage for the
comfortable guitar tuning. The visual peaks must be symmetrical at the
middle position of the ôTuneö control slider.
The built-in sequencer allows opening and playing MIDI (*.mid)
and RIFF MIDI (*.rmi) files. New tracks can be also added to the open files.
A new record in the sequencer is made by addition of a new track. Thus,
MIDI record can be created from several tracks. The record can be saved in
MIDI or RIFF MIDI files.
The software provides the selection of the input audio device and the
output MIDI device. Due to the selection option of the output MIDI device,
AudioToMidi is able to operate with an external software sequencer. The
driver Sonic Foundry Virtual MIDI Router (VMR) is recommended for this
purpose. The distribution conditions of this file with the description in
English are found in the enclosed file Sonic Foundry MIDI Router.wri (the
driver package is not included). MIDI signal, both real time and written to
the built-in sequencer, can be applied to the external software sequencer. In
the latter case all tracks are combined into one track.
The audio *.wav, *.mp3, *.au files and the like can be converted by
the playback by the appropriate software giving the sound to the audio
device selected at the AudioToMidi input (usually Wave Mapper).
2.0 MINIMUM SYSTEM REQUIREMENTS
Processor: P75.
OS: Windows 95 or Windows NT.
Memory: Approximately 1MB free ROM.
Hard disk: Approximately 1MB free space.
Devices: Any sound card, which is not worse than
SB16 in possibility.
Drivers and application programs of the sound
card must be installed to OS.
Note: No sound is given to PC speaker under
Windows NT, the appropriate option is
disabled.
3.0 INSTALLATION
This software product is installed by simple copying of files to any
directory preserving the archive catalog tree structure. No additional
libraries or drivers are required. The file AudioToMidi.exe runs the
software.
4.0 OPERATION PRINCIPLE
A continuous sample flow representing a digitized sound from the
ôWave Inö list device is given to the program input. This signal is given to
the massive of sensors; each tuned at a particular frequency. This frequency
is equal the frequency of the note associated with the sensor, at some
possible correction. The value representing the sound intensity within the
domain of the sensor own frequency is generated at the each sensor output.
These values are graphically presented in the ôSpectrumö window.
The note frequency is commonly calculated. Note A of the first
octave has the frequency of 440 Hz. When the note is raised or lowered at
1/2 tone, the frequency is multiplied at or divided by the value, equal to the
12-power root of two. Hence, if the note is raised or lowered at 12 semi-
tones, i.e., at one octave, the frequency is multiplied at or divided by 2. A of
small octave matches 220 Hz, A of large octave - 110 Hz, A of the second
octave - 880 Hz, A of the third octave - 1760 Hz, etc. The correction of the
sensor frequency is a function of the position of ôTuneö slider. At the
middle position of the slider the correction is equal to zero. At the most left
position the correction makes the frequency match the note frequency,
which is 1/2 tone below the note associated with the sensor. At the most
right position the frequency matches the frequency of the note which is 1/2
tone above the note associated with the sensor. If the slider is moved
smoothly from the most left to the most right position, the frequency of each
sensor is also smoothly changed from the lowest to the highest value.
Sensor sensitivity individually set by the ôEqualizerö control.
ôSensitivityö control changes the sensitivity of all sensors simultaneously.
The sensor sensitivity is increased when moving the ôEqualizerö or
ôSensitivityö slider upward.
The selectivity of sensors set by the ôSelectivityö control is
graphically represented in the respective window. The plot shows the sensor
selectivity as a function of the audio wave frequency. The middle vertical
line corresponds to the sensor own frequency. The adjacent vertical lines
correspond to the frequency values differentiated by the 12-power root of
two from the sensor own frequency. The frequency by the abscissa axis
grows from left to right by the logarithmic scale. The selectivity plot is
almost symmetric with the maximum at the own frequency.
The sensors are characterized by some inertia, displaying the rate of
reaction on the appearance or disappearance of the sound in the own
frequency domain. The selectivity and inertia of the sensor strictly depends
one from another. The better is the sensor selectivity (the narrowest
selectivity plot), the more this sensor is inert, i.e. slow. The optimum
selectivity value is experimentally chosen being dependent from a particular
converted melody, its, tempo, polyphony, percussion, average note duration,
etc.
The values generated at the sensorsÆ outputs are periodically scanned
to detect the peaks. The peak is the sensor with the output value above the
values of both adjacent sensors. The scanning is performed from left to
right, i.e., from the low note sensor to the high note sensor. Before the start
of each scanning cycle a uniform threshold value is set for all sensors. This
threshold value depending upon the position of the ôGateö slider is
presented by a horizontal dotted line in the ôSpectrumö window.
When a peak is detected, the addition is made to the threshold value
of the sensors presenting the notes above the current note at 12, 19 and 24
semi-tones. The addition to the threshold depends upon the peak sound
intensity and the histogram shape in the ôHarmonic M