Audio files, like all other computer data files, are stored in a binary table which is a series of zeros and ones representing data bits. To distinguish file type and to provide instructions to audio software, most sound files begin with a block of data called the file header. Some audio file are also "wrapped" or "tagged" to provide additional features or licensing information. File extensions vary by file type and by the program that created or uses them. Unfortunately it has become common practice in digital audio to use such a wide variety of file extensions for essentially the same formats that in many cases there is no way to know the actual file format or the codec short of opening the file.

When you double click on a sound file your Windows operating system first reads the file extension and then calls the appropriate executable program (EXE), usually a player, to open the file. The executable program first reads the source file header to determine how to interpret the binary data such as the compression method, the sampling rate and other properties that the program needs in order to play or otherwise manipulate the contents. Some software producers use proprietary file extensions for common file types so you cannot always tell the file's format from the file extension. See Opens with... for more.

When the program opens a specific sound file, if it finds that the file header is valid and that the audio data matches the expected criteria, the data will be decompressed if necessary, and then loaded into memory where it can be played or manipulated.

The illustration below represents 3 samples of 8-bit PCM audio which are the zeros and one's that we mentioned at the beginning of this section.

It is important to note here that the higher the sample rate (the number of bits in each second of sound) the higher the fidelity of playback and the higher the disk space requirement. Thus there's always a trade-off between quality and resources.

The following audio formats are directly supported within Fx Audio Editor. We'll talk briefly about each.

• Uncompressed WAV PCM

• Compressed WAV (ADPCM, GSM, DSP and others)

• MP2 (MPEG 1/2 Layer-2)

• MP3 (MPEG 1/2 Layer-3)

• VOX (Dialogic ADPCM)

• WMA (Windows Media Audio 9)

• RAW audio (PCM, A-LAW, U-LAW)

• MPC (MusicPack)

• AVI (audio track)

• Ogg Vorbis (Version 1.0)

• G.721, G.723, G.726

• AIFF, AIF (Apple audio format)

• AU (Sun UNIX audio format).

• CD audio (CDDA)

Uncompressed WAV PCM (Pulse Code Modulation) is the primary, generic method used for storing uncompressed digital audio. PCM is used in CD Audio and DAT. Typical file extensions are WAV, AIF, AIFF and PCM. As we saw in the illustration above, PCM is a one for one representation of pure binary data.

DPCM (Differential Pulse Code Modulation) is a form of compression that stores the value of the first sample as a benchmark and then stores the differences between each consecutive sample. Because DCPM compression uses only 4 bits to store the difference rather than the 8 bits required for PCM data, an 8-bit PCM file is compressed at 2:1 resulting in a file half the original size. When applied to16-bit PCM the result is a 4:1 compression ratio. This type of compression is known as lossy meaning that some quality is lost in the process.

ADPCM (Adaptive Differential Pulse Code Modulation) uses an algorithm to analyze blocks of samples then stores the first bit of major changes and a key that can predict the values of a number of the following samples. Compression is quite good when the audio in a particular file fits the expected criteria but poor when the sound in the file is very complex.

MPEG, MP2, MP3, (ISO/CCITT Moving Pictures Expert Group JTC1/SC29/WG11) the standard includes MPEG-1 or MPEG-2 Layer 2 and Layer 3 audio more commonly known as MP2, MP3 as well as AAC Audio. MPEG compression is based upon the concept of removing all sound that cannot be heard by the human ear to include overlaps where one sound cancels another. The compression ratio is excellent with little reduction in quality.

VOX (Dialogic ADPCM) format is optimized for human voice and is widely in telephony applications. Compression is 4:1, all files are mono with no header so they must be saved using the VOX file extension.

WMA (Windows Media Audio) has gone through several iterations. File are roughly equivalent to MP3 of arguably equal or better quality. WMA files are also wrapped with extended tag information and a very comprehensive licensing system used by Wal-Mart, Microsoft and other on-line music retailers. Licensed WMA music files will raise a Windows file access error in most players that support WMA playback.

To edit unsupported formats the files must be played in their respective players and recorded in Fx Audio Editor (see Opens with...) and saved to a new file name. All tags and wrapper information will be lost by this process.