The SoundStreamHead tags define a sound effect which is to be loaded with a set of SoundStreamBlock tags. It defines the sound once and for all.
Streaming sound has a strong side effect when playing a movie: it will force a synchronization between the images and the audio. Thus some images may be dropped if the drawing isn't fast enough.
Streaming sound effects should be either used in the main movie or in a sprite which needs a sound track properly synchronized to the sprite animation. Otherwise, it is much better to use the DefineSound and StartSound tags.
It seems (though it isn't documented) that using a SoundStreamHead tag by itself (without any sound blocks) is taken as a hint of how to play all the other sounds in an animation (see the f_playback_rate)
Important note: there can be only one streaming sound per movie clip including the main movie. If you need multiple sound effects or music to be played back, these will have to be merged at the time you create the movie in a single sound which will then be played back as a single sound track.
Some of the fields in the SoundStreamHead tag can't have all the possible values when defined in an older version of SWF. What follows describes each field in more details.
The f_playback_rate and f_sound_rate define the rate at which the data should be played and the exact rate of the data found in the SWF file.
rate = 5512.5 * 2 ** f_playback_rate rate = 5512.5 * 2 ** f_sound_rate
The f_playback_size and f_sound_size define the number of bits found in the data (0 for 8 bits and 1 for 16 bits.) Note that with a SoundStreamHead tag (18) only 16 bits data are allowed (both are always set to 1). If the compression mode isn't Raw or Uncompressed then the f_sound_size must be set to 1.
The f_playback_stereo and f_sound_stereo define whether the sound should be played on both speakers and whether the data includes both channels.
The f_compression entry defines the compression mode. The list of sound compression modes can be found with the DefineSound tag. Note that with a SoundStreamHead tag (18) only the ADPCM compression is accepted. All the compression modes are accepted in a SoundStreamHead2 tag (45).
The f_sample_size represents the average number of samples per frame. It is used to ensure a proper synchronization. Note that there can be more (and even less in MP3) samples within a given frame.
The f_latency_seek is usually zero. It exists only when the MP3 compression mode is used. It defines the number of samples to skip at the beginning of the very first frame (I'm not totally sure what this is to tell you the truth... I'll tell you more once I know more).