TS-AudioToMIDI is a real-time converter. This means you can use this program as MIDI sound generator for non-MIDI instruments such as guitar or flute. TS-AudioToMIDI can detect notes played by any instrument in real-time and play it synchronously or store as MIDI command. Hence one instrument can be replaced with another, for example guitar sound can be replaced with violin or organ timbre.
To start real-time transcription first select the source for the sound you are going to transcribe. You can do this with the help of "Wave In" control at the top of main window. By default "Wave Mapper" is selected, as it fits for most systems. At the right of "Wave In" control "Start/Stop Wave In" button is located. It turns sound sensing on/off . To use microphone as sound source, click "Windows Mixer" button near "Start/Stop Wave In". Standard Windows volume control will appear. Choose "Options"->"Properties" menu item and having selected "Recording" in "Adjust volume for" section, press OK. In appeared "Recording control" dialog select "Microphone" (or any other required sound source) input device and adjust volume for it.
Having returned to TS-AudioToMIDI, make sure that "Start/Stop Wave In" is on and take a look at the spectrum Window. The volume of each spectral component is monitored here in real-time (if you select microphone as input device for your instrument or sing something into it, you will see the picture changing dynamically). Recognition process starts as soon as you turn on "Start/Stop Wave In" button.
As a rule, without adjusting Threshold and Noise Gate controls either all notes are rejected as too silent or too much odd notes are passed, producing "mess" effect. To avoid both this extremes verify sound level. All algorithms except Mono Correlator have four controls affecting minimum note level, they are Threshold, Noise Gate, Selectivity and Sharpness. First play the record you are going to recognize and look at it's level in Spectrum window. Yellow dotted line in it can be moved by means of Threshold slider at the right. This line represents minimum sound level that is analyzed for transcription, and all peaks lower it are ignored. You can also use Noise Gate control to change amplification level. Selectivity and Sharpness set sensors parameters, i.e. control parameters of oscillatory circuits assigned for detection of spectral peaks. This influence the shape of spectral peak produced by sensor and hence determines whether this peak will be detected as a part of musical note or not. You can change all parameters during real-time recognition process as well.
"Play/Keep silence" control in the lower right corner allows choosing sound output method. When choosing "MIDI Device", the device you have selected in "MIDI Out" control will be used ("MIDI Mapper" by default). After turning "MIDI Device" on you will hear the sound transcribed in real-time, simultaneously with initial sound. By default transcription result is not stored. You can store it by using "Record" or "Auto Record" button at MIDI sequencer. "Record" button starts recording as soon as it is pressed in contrast to "Auto Record" button that starts recording as soon as first note is transcribed. When recording is on, you can turn "MIDI Device" both on or off, this will not influence the notes recorded.
Please, note that acoustic feedback effect during real-time recognition may lead to incorrect recognition if you are using omnidirectional microphone. When you turn on MIDI sound during recognition, this sound is received back by microphone and goes to Wave In once more. This may lead to "sealed" notes. To solve this problem either decrease output volume or use directional microphone.
Related topics:
How TS-AudioToMIDI recognizes music
Recommendations on improving recognition quality
|
