You can use TS-AudioToMIDI both for recognizing pre-recorded audio files and real-time recognition. Both ways have they own advantages and disadvantages. Let's dwell on them in detail.
This mode allows recognizing sound directly from microphone, line input or any other input channel of your sound card or other installed device. Full list of available devices can be found in "Wave In" list, sound card input channel can be selected from standard Windows "Recording Controls" dialog. Input sound is not recorded. When you are, say, singing to microphone with TS-AudioToMIDI on-fly recognition mode active, only recognized notes are stored into build-in MIDI sequencer, and the sound of your voice itself is lost. This allows saving disk space, cause digital audio files, even compressed, take hundreds times more disk space than MIDI. On the other hand, optimizing recognition settings usually requires several tries, and it is hardly possible to sing or play the musical instrument absolutely the same thing several times. However, when you are regularly using the same instrument and microphone with constant distance between them, optimal recognition settings differ slightly.
In this mode TS-AudioToMIDI can be used as a music processor which allows some non-MIDI instrument (like a guitar) to sound as, say, a violin or a piano or whatever instrument you've got in your MIDI synthesizer.
Recognizing pre-recorded sound.
You can recognize any sound file in any of the supported formats. Advantages are: input data are always the same which allows adjusting recognition settings with greater accuracy; recognition process goes faster than when on-fly and you don't need to sing or play the instrument several times. But it is not always possible to tune up Harmonic model, for that requires recording of a separate note played with the same instrument as the whole melody. Also this method requires some disk space to store the initial audio file.
Performing recognition of pre-recorded audio