AST file
AST files contain streamed audio data, and are used in Pikmin 2. To simply play audio files, you can use vgmstream, an audio playback library for multiple video game specific audio formats.
To convert .wav to .ast, you can use JATAST. This defaults to ADPCM encoding, as the GameCube's disc reading (whether on real hardware or emulated with dolphin) will not be able to keep up with large streamed music otherwise.
File Format[edit]
Header[edit]
The file starts with a 0x40 bytes long header as follows:
| Offset | Type | Name | Description |
|---|---|---|---|
| 0x00 | int32 | magic | 'STRM' 0x5354524D |
| 0x04 | int32 | size | (File Size - 0x40) |
| 0x08 | u16 | Format | 0 = ADPCM, 1 = PCM16. |
| 0x0A | u16 | BitsPerFrame | Usually 16. |
| 0x0C | u16 | channelCount | Number of channels. |
| 0x0E | bool (u16) | loop | 0xFFFF = true, 0x0000 = false |
| 0x10 | int32 | sampleRate | Sample Rate. |
| 0x14 | int32 | sampleCount | Sample Count |
| 0x18 | int32 | loopStart | Start of loop in samples (Aligned to nearest 16 samples) |
| 0x1C | int32 | loopEnd | End of loop, aligned to nearest 16 samples. SampleCount if no loop. |
| 0x20 | int32 | blockSize | Block size. Generally 0x00002760.
NOTE: This is the size of ONE CHANNEL. If you have 2 channels, the block will be 4EC0 in length, but this should still say 0x2760 |
| 0x24 | int32 | unknown | 0 |
| 0x28 | u8 | volume | Volume modifier, byte. 0x7F = 100% |
| 0x29 | Byte[0x17] | padding | Padding to align header to 0x40 |
| 0x40 | BLCK[] | ... | ... |
Block / BLCK[edit]
| Offset | Type | Name | Description |
|---|---|---|---|
| +0x00 | int32 | magic | 'BLCK' 0x424C434B |
| +0x04 | int32 | size | Block size, 0x2760 if full block.
NOTE: See the above comment regarding block size. This will be the length of one channel. The last BLCK's size might differ, but the rest will generally remain 0x2760 |
| +0x08 | u16[2][6] | sampleHist | Historical samples for ADPCM4 only. |
| +0x20 | u8[size][channelCount] | sampleData | Sample data. Each channel will be packed directly following the previous channel. |
| +(size + 0x20) | Next BLCK |
Trivia[edit]
When encoding the sampleHist part of the block in ADPCM, this will be the historical values (last = l, p = penultimate)
Each set of two u16's represents one channel in the AST. If there aren't that many channels, the value will be set to 0.
The value of these will be the last evaluated (decoded) samples from the previous block (matching channel). Without these, the predictor for the current block will overscale, resulting in "crunchy" Audio.