A MATLAB-based implementation of an Automatic Music Transcription (AMT) system wasdeveloped.

The goal is to transcribe a monophonic music audio signal into symbolic music notation (MIDI numbers). This involves:

• Displaying the time-domain waveform and spectrogram (STFT) of the input audio.

• Detecting the onset and offset times of each musical note.

• Estimating the corresponding MIDI note number. (

  MIDI_number = 69 + 12 * log2(freq / 440)
  

  where freq is the detected frequency in Hz, and 440 Hz corresponds to the A4 note.)

• Groups temporal note segments based on continuity.

• Visualizing both the detected and the ground-truth MIDI notes.

• Saving the output as a .csv file.

• Uses Short-Time Fourier Transform (STFT) for time-frequency analysis.
• Converts spectral peaks to MIDI note numbers.
• Groups temporal note segments based on continuity.
• Compares detected notes with ground-truth annotations visually and quantitatively.
• Saves results to detected_notes.csv.