WAV Format

From ModdingWiki
Jump to navigation Jump to search
WAV Format
Format typeSound
HardwarePCM
Number of sounds1
Sampling rateAny
Channel countAny
Bits per sample8,16,24,32
Compressed?No
Tags?Name
Games

The WAVE File Format is a file format for storing digital audio (waveform) data. It supports a variety of bit resolutions, sample rates and channels. The format is a subset of Resource Interchange File Format (RIFF) created by Microsoft, which itself inspired by the Interchange File Format (IFF).

Data in the WAVE format is stored in chunks. Each chunk begins with a 4 character chunkId which is often a recognizable ASCII sequence, and is followed by a 32-bit chunk length, and then the payload for that chunk.

General Layout

In general there is no restriction on the order of chunks in a WAVE file, except that the format chunk must precede the data chunk.

The minimum layout would be


Type Name Description
FOURCC riffID RIFF.
UINT32LE fileLength Length of the file (less the 8 bytes for the RIFF header and length).
FOURCC formID WAVE identifies this as a WAVE file.
FOURCC chunkID Type of chunk.
UINT32LE chunkLength Length of the chunk (less the 8 bytes for the chunkID and length).
... etc More chunks until the file is complete.

Format Chunk

The format chunk describes way that the sample data is organised in the WAVE file.

Type Name Description
FOURCC chunkID 'fmt '.
UINT32LE chunkLength Length of the chunk.
INT16LE formatTag Indication of any compression. If 1 there is no compression. If compression is used, there may be additional fields in the chunk.
UINT16LE channels Number of channels in the file.
UINT32LE samplesPerSec Sample rate in Hz.
UINT32LE avgBytesPerSec Average bytes per second. (Useful for calculating appropriate buffer sizes).
UINT16LE blockAlign should equal channels * ((bitsPerSample / 8) rounded up)
UINT16LE bitsPerSample number of bits per sample.

Data Chunk

The data chunk contains the sample data.

PCM data with a resolution greater than 8 bits per sample is represented in twos-complement form. i.e 16-bit sample data ranges from -32768 to +32767, the zero-crossing point is at 0.

If the resolution is 8 bits or less, then the data is represented within an unsigned byte. The full dynamic range is from 0 to 255 and the zero-crossing point is at 128.

If the number of bits per channel is not divisible by 8, the sample is left justified within the binary field. i.e. only the most significant bits are used; the unused least significant bits should be zeroed.

If the number of bits per channel is more than 8, the data is stored in a little-endian fashion.

If the number of channels is more than 1, then the channels are interleaved. The first sample is stored for each of the channels, then the second sample for each of the channels etc.

There are some standard channel orders.

stereo: Left, Right

3 channel: Left, Right, Center

4 channel: Left, Center, Right, Surround

Quad: Front Left, Front Right, Rear Left, Rear Right

6 channel: Left Center, Left, Center, Right Center, Right, Surround

Type Name Description
FOURCC chunkID 'data'.
UINT32LE chunkLength Length of the chunk.
UINT8[chunkLength] waveFormData Sample data packed as described above
UINT8 paddingByte May or may not be present. The length of data should be an even number of bytes. A single padding byte may be needed.

Other Chunks

Other optional chunks may appear in the file.

FourCC Name Description
'cue ' Cue Cue points
'plst' Playlist The playlist specifies an order for a series of cue points. It can indicate looping sections.
'LIST' IFF List An IFF List embedded in the file. Most often for an adtl.
'adtl' Associated Data List A list of data associated with the cue points
'smpl' Sampler Basic parameters that a sampler could use to play the waveform e.g. information about looping for a sustain effect

Credits

This file format was documented by Val Mushinskiy. If you find this information helpful in a project you're working on, please give credit where credit is due. (A link back to this wiki would be nice too!)