DRO

DOSBox Raw OPL or DRO is the format used by DOSBox to capture OPL output. The format uses a clear header but went through two incompatible versions. The first version was used up to DOSBox 0.72, the second version replaced it starting with DOSBox 0.73.

Technical information

All numerical values are little-endian.

Offset	Size	Description
0	char[8]	Signature (must be `DBRAWOPL`)
8	uint16	First version word
10	uint16	Second version word

Version number identification is complicated by the fact the version words aren't read in the same order depending on the specifications. In the 1.0 version of the format, the minor version number comes first, and the major version number is the second word. So for version 1.0, the value is 00 00 01 00. However, version 2.0 of the specifications reverse the order, so for version 2.0, the value is 02 00 00 00. For this reason, the 1.0 version also became known as version 0.1.

Note: Some files using the old version format contain large garbage values in the minor version word. ZDoom will refuse to read them. You will need to set the minor version to 00 00 with a hex editor.

Delay values are indicated in milliseconds - 1. Since a delay of 0 is deemed to be useless, the byte value is increased by one to get the actual delay.

The header is then followed by a subheader. The two different formats have completely different subheaders.

Format 1.0

Size	Description
12	uint32	Song length (in millisecond)
16	uint32	Song data size (in bytes)
20	uint8	Hardware type (0: OPL2, 1: OPL3, 2: Dual OPL2)
21	byte[3]	Optional null padding (may not be present: if non-null, these bytes are song data)

Format 2.0

Size	Description
12	uint32	Song data size (in pairs of bytes)
16	uint32	Song data length (in milliseconds)
20	uint8	Hardware type (0: OPL2, 1: Dual OPL2, 2: OPL3)
21	uint8	Data format (ZDoom only supports data format 0)
22	uint8	Compression method (ZDoom only supports compression 0, meaning no compression)
23	uint8	Short delay code (command value used to introduce a one-byte delay, should always be 122)
24	uint8	Long delay code (command value used to introduce a one-byte delay multiplied by 256, should always be 123)
25	uint8	Codemap table size (must be 126 or less, technically should always be 122)
26	uint8*	Codemap table (associate OPL registers with index values)

The standard codemap looks like this:

Index	Register	Index	Register	Index	Register	Index	Register	Index	Register	Index	Register	Index	Register	Index	Register
0	01	1	04	2	05	3	08	4	BD	5	20	6	40	7	60
8	80	9	E0	10	21	11	41	12	61	13	81	14	E1	15	22
16	42	17	62	18	82	19	E2	20	23	21	43	22	63	23	83
24	E3	25	24	26	44	27	64	28	84	29	E4	30	25	31	45
32	65	33	85	34	E5	35	28	36	48	37	68	38	88	39	E8
40	29	41	49	42	69	43	89	44	E9	45	2A	46	4A	47	6A
48	8A	49	EA	50	2B	51	4B	52	6B	53	8B	54	EB	55	2C
56	4C	57	6C	58	8C	59	EC	60	2D	61	4D	62	6D	63	8D
64	ED	65	30	66	50	67	70	68	90	69	F0	70	31	71	51
72	71	73	91	74	F1	75	32	76	52	77	72	78	92	79	F2
80	33	81	53	82	73	83	93	84	F3	85	34	86	54	87	74
88	94	89	F4	90	35	91	55	92	75	93	95	94	F5	95	A0
96	B0	97	C0	98	A1	99	B1	100	C1	101	A2	102	B2	103	C2
104	A3	105	B3	106	C3	107	A4	108	B4	109	C4	110	A5	111	B5
112	C5	113	A6	114	B6	115	C6	116	A7	117	B7	118	C7	119	A8
120	B8	121	C8	122	short delay					123	long delay

External links

This article is issued from Zdoom. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.