TSDL_AudioCVT = record
needed: Integer; // Set to 1 if conversion possible
src_format: UInt16; // Source audio format
dst_format: UInt16; // Target audio format
rate_incr: double; // Rate conversion increment
buf: PUInt8; // Buffer to hold entire audio data
len: Integer; // Length of original audio buffer
len_cvt: Integer; // Length of converted audio buffer
len_mult: Integer; // buffer must be len*len_mult big
len_ratio: double; // Given len, final size is len*len_ratio
filter_index: Integer; // Current audio conversion function
|needed||Set to one if the conversion is possible|
|src_format||Audio format of the source|
|dest_format||Audio format of the destination|
|rate_incr||Rate conversion increment|
|len||Length of the original audio buffer in bytes|
|len_cvt||Length of converted audio buffer in bytes (calculated)|
|len_mult||buf must be len* len_mult bytes in size(calculated)|
|len_ratio||Final audio size is len*len_ratio|
|filters(..)||Pointers to functions needed for this conversion|
|filter_index||Current conversion function|
The TSDL_AudioCVT is used to convert audio data between different formats. A TSDL_AudioCVT structure is created with the SDL_BuildAudioCVT function, while the actual conversion is done by the SDL_ConvertAudio function.
Many of the fields in the TSDL_AudioCVT structure should be considered private and their function will not be discussed here.
This points to the audio data that will be used in the conversion. It is both the source and the destination, which means the converted audio data overwrites the original data. It also means that the converted data may be larger than the original data (if you were converting from 8-bit to 16-bit, for instance), so you must ensure buf is large enough. See below.
This is the length of the original audio data in bytes.
As explained above, the audio buffer needs to be big enough to store the converted data, which may be bigger than the original audio data. The length of buf should be len*len_mult.
When you have finished converting your audio data, you need to know how much of your audio buffer is valid. len*len_ratio is the size of the converted audio data in bytes. This is very similar to len_mult, however when the convert audio data is shorter than the original len_mult would be 1. len_ratio, on the other hand, would be a fractional number between 0 and 1.