MPEG-1 Audio Layer III or MPEG-2 Audio Layer III, commonly known as MP3 is a proprietary compression format for digital audio which uses an algorithm to achieve loss smaller file size. It is a common audio format used for music both on computers and portable audio players.

• The MPEG-1 files correspond to sampling rates of 32, 44.1 and 48 kHz.
• The MPEG-2 files correspond to sampling rates of 16, 22.05 and 24 kHz.

Technical details

In this layer exist several differences from MPEG-1 and MPEG-2 standards, among which is the so-called filter bank to have greater design complexity. This improved frequency resolution worsens temporal resolution, introducing pre-echo problems that are predicted and corrected. It also allows audio quality as low as 64 kbps rates.

Filter Bank

The filter bank used in this layer is called polyphase bank hybrid filters. It is responsible for conducting the mapping from the time domain to the frequency for both the encoder and to the decoder reconstruction filters.

The psychoacoustic model

Compression is based on reducing the irrelevant dynamic range, that is, the inability of the human hearing system to detect quantization errors in masking conditions. This standard divides the signal into frequency bands that approximate the critical bands, and then quantizes each sub-band based on the noise detection threshold within that band.

Digital encoding and quantification

The solution proposed by this standard for the distribution of bits or noise, is done in an iteration cycle consisting of an internal and an external cycle. Examines both the output samples of the filter bank as SMR (signal-to-mask ratio) provided by the psychoacoustic model, and adjusts the bit allocation or quantization noise, according to the scheme used to meet the requirements of bit rate masking and simultaneously.

Packaging or bitstream formatter

This block takes the quantized samples of the filter bank, along with mapping data bits/noise and stores the encoded audio and some additional data frames.

Volume

Volume normalization, also known as Audio normalization, basically consists of leveling the volume of the tracks that make an album, allowing you to listen to songs at the same volume always.

Structure of an MP3 file

The construction of an MP3 file is with different frames which in turn consist of a header and data itself. This data stream is the “elementary stream”. Each of the frames is independent, ie, you can cut frames of an MP3 file and then play on any MP3 player. The header consists of a sync word which is used to indicate the beginning of a valid frame. Following are a series of bits indicating that the file is analyzed Standard MPEG file and whether or not you use the layer 3. After all this, the values differ depending on the type of MP3 file. The value ranges are defined in ISO / IEC 11172-3 standard.

Bit rate

The bit rate of an MP3 indicates the density of the audio information contained in the file. Therefore, every second in a MP3 with 192kbps contains 192 kilobits (24 bytes) of data. The higher the bit rate, the more information contained in the MP3; the more information it contains, the quality is closer to the original audio recording. The bit rate also determines the size of the MP3 file: A four-minute song encoded at 128 kbps occupies a little over 3.5 MB, while the same song encoded at 320 kbps requires more than 9MB of space.

Variable bit rate

Encoding with a variable bit rate changes the bit rate of the MP3 depending on the type of data present in a particular point  of  the audio file. For example, an encoder with variable bit rate could encode a portion of information at 320 kbps; however, when the audio contains a section of silence encoder low bit rate up to 32 kbps. If the bit rate of an MP3 has an unusual number (for example, 204 kbps), this indicates that the creator of the MP3 encoded with a variable bit rate.

Conversion bitrates

Although most conversion programs can modify an MP3 to a different bit rate, this is not a good idea. Convert an MP3 to a higher bit rate does not add audio information to the file, so the sound quality is not increased. Convert an MP3 to a lower bit rate reduces the file size, but may also introduce audio artifacts, since the encoder is compressing a file that is already compressed. If you need that an MP3 file has a different bit rate, re-encode an MP3 from the original audio CD or WAV file.