Normally when mixing channels, the source channels are sent to the output in the same order; the left input is sent to the left output, and so on. Sometimes something a bit more complex than that is required. For example, if the source has more channels than the output, you may want to "downmix" the source so that all channels are present in the output. Equally, if the source has fewer channels than the output, you may want to "upmix" it so that all output channels have sound. Or you may just want to rearrange the channels. Matrix mixing allows all of these.
A matrix mixer is created on a per-source basis (you can mix'n'match normal and matrix mixing), by using the BASS_MIXER_CHAN_MATRIX and/or BASS_MIXER_CHAN_DOWNMIX flag when calling BASS_Mixer_StreamAddChannel or BASS_Mixer_StreamAddChannelEx. The matrix itself is a 2-dimensional array of floating-point mixing levels, with the source channels on one axis, and the output channels on the other. Some simple examples are shown below.
When using matrix mixing, the source channel's volume attribute still has effect, but the pan attribute does not. Whenever necessary, panning changes can be achieved by modifying the matrix.
float matrix[2][2] = {
{1, 0}, // left out = left in
{0, 1} // right out = right in
};
BASS_Mixer_ChannelSetMatrix(handle, matrix); // apply the matrix
In = stereo, Out = swapped stereo.
float matrix[2][2] = {
{0, 1}, // left out = right in
{1, 0} // right out = left in
};
BASS_Mixer_ChannelSetMatrix(handle, matrix); // apply the matrix
In = stereo, Out = mono.
float matrix[1][2] = {
{0.5, 0.5} // mono out = half left + right in
};
BASS_Mixer_ChannelSetMatrix(handle, matrix); // apply the matrix
In = stereo, Out = quadraphonic (4 channels).
float matrix[4][2] = {
{1, 0}, // front left out = left in
{0, 1}, // front right out = right in
{1, 0}, // rear left out = left in
{0, 1} // rear right out = right in
};
BASS_Mixer_ChannelSetMatrix(handle, matrix); // apply the matrix