Home
TS-AudioToMIDI Web Online Help
Introduction
What is Music Recognition
TS-AudioToMIDI in brief
What's new?
Features
Features
Supported formats
System Requirements
Installation
Installing and Uninstalling TS-Audio2MIDI
Quick Start
How to transcribe a WAVE file
How to transcribe music in realtime
Using TS-Audio To MIDI
Basic Operation
Wave Recording
Perform recognition of pre-recorded audio
Realtime recognition
MIDI Playback
Audio Playback
Advanced Topics
Setting Equalizer
Tuning up Selectivity Window
Setting Recognition Parameters
Choosing Recognition Algorithm
Auto Tune
Setting up Threshold and Noise Gate
Setting Harmonic Model
Saving settings
TS-Audio To MIDI Reference
TS-AudioToMIDI Main Window
Wave Playback and Convert controls
Wave Recorder
Device Controls
Tune
Spectrum Analyzer and Keyboard
Filter Window
Graphic Equalizer
Selectivity Window
Noise Gate and Threshold
Instrument selector
Transponse control
Volume control
Harmonic model
MIDI Settings dialog
Algorithm selector
MIDI Channel selector
Minimal Note and Pause duration
Play/Keep silence control
Build-in MIDI Sequencer
Save and Load Recognition Settings
Time Window
MIDI Player position
Spectrum Window
Additional Info
How does TS Audio to MIDI recognizes music
Recognizing pre-recordered files vs on-fly recognition
Recomendations on improving recognition quality
Contacts & Support
Registration
License agreement
FAQ & Troubleshooting
What is Music Recognition
   
TS-AudioToMIDI performs recognition of music audio files and allows saving the results as MIDI files.
Each melody can be represented in two distinctly different forms.
The first is notes representation, when we describe melody as a set of commands such as "play note A with violin, volume equals to 100, duration is 400 ms". Both MIDI sequencer and symphony orchestra can be controlled by such a set of commands. Two most common ways to store a melody described in this way are MIDI file and musical notation.
The other way is to store waveform. Waveform is the dependence of acoustical pressure on time. Not only music can be stored as waveform, hence not any waveform can be converted to notes. For example human speech can't. Analog tape and wave format file store sound in this way. During last decade numerous waveform-conversion algorithms were developed, both with and without data loss. Most common formats of compressed waveform are mp3, MPEG, RealAudio(.ra file extension), and others.

Music recognition is what musician does when playing music by ear and then recording it to staves. Recognition requires previous knowledge of music structure, common instruments etc. and ability to detect musical sounds with sufficient precision. In most cases music heard cannot be recognized unambiguously because the same waveform can be interpreted as several different note sets. This is very usual for polyphonic music when even professional musician can hardly say what notes were played by each particular instrument.

TS-AudioToMIDI performs automated music recognition. This program analyzes input waveform and tries to detect what notes were played. Some basic knowledge about music structure, such as Harmonic Model are build-in into recognition algorithm, an array of sensors with controlled selectivity and sensitivity allows precise detection of spectral peaks. However science still hasn't developed an errorless algorithm for recognition, so mistakes are possible. Adjusting settings individually for each recording allows reducing error rate. In general any recognition process looks as conversion from WAVE info MIDI format.

Please note that TS-AudioToMIDI does not convert MIDI to WAV. This task is the opposite and is performed by programs called software MIDI renders.

Related topics:
TS-AudioToMIDI in brief