1. Audio Decoding

The WAV file is decoded from its PCM format into raw audio samples. This involves reading the audio data header to understand the sample rate, bit depth, and number of channels.

2. Psychoacoustic Analysis

The encoder performs a Fast Fourier Transform (FFT) to analyze the frequency content of the audio. It identifies sounds that are less perceptible to human hearing based on:

• Frequency masking (louder sounds mask quieter ones at similar frequencies)
• Temporal masking (sounds mask other sounds that occur shortly before or after)
• Absolute threshold of hearing (sounds too quiet to be heard)

3. MDCT Transformation

The audio is divided into frames and transformed using Modified Discrete Cosine Transform (MDCT). This converts the time-domain audio samples into frequency-domain coefficients.

4. Quantization

Based on the psychoacoustic model, the encoder allocates bits to different frequency bands. More bits are given to perceptually important frequencies, while less important frequencies are coarsely quantized or discarded.

5. Huffman Coding

The quantized coefficients are compressed using Huffman coding, which assigns shorter codes to more frequent values.

6. Frame Packing

The processed data is packed into MP3 frames with headers containing metadata like bitrate and sampling rate. Each frame is 26ms of audio (1152 samples at 44.1kHz).