das15@cunixa.cc.columbia.edu (Douglas A Scott) (02/26/91)
Hi. Would someone kindly explain how mu-law encoding works in the 8-bit NeXT soundfiles? I am trying to expand a soundfile-editing program to allow it to read and display that type of file (it currently reads 16- and 32-bit soundfiles). The program I have is not native to the NeXT, so it does not automatically handle such encoded files. Thank you very much for any information. ___________________________________________________________________________ Douglas Scott zardoz!doug%woof.columbia.edu
eps@toaster.SFSU.EDU (Eric P. Scott) (02/26/91)
mu-law is a logarithmic transformation that provides approximately the same fidelity in 8 bits as 12-13 bits of linear representation would. Treat it as compressed data; don't do any math on it. (Specifics of mu-law and A-law encoding can be found in telecommunications texts; they are standardized.) In 16-bit linear, you deal with signed integer values. 0 is zero, and the range is +-32767. mu-law data is stored in an inverted sign+magnitude form: +-+-+-+-+-+-+-+-+ | | | +-+-+-+-+-+-+-+-+ ^ level; all ones is zero amplitude; all zeros is maximum | 0 = negative 1 = positive The levels are nonlinear; there is finer resolution at the low end. NeXT provides two functions to convert between 16-bit linear and 8-bit mu-law: #import <sound/convertsound.h> unsigned char SNDMulaw(short) /* linear to mu-law */ short SNDiMulaw(unsigned char) /* mu-law to linear */ To "mix" two mu-law sounds, you have to do something like: SNDMulaw(SNDiMulaw(one)+SNDiMulaw(two)) On SunOS 4.1 and later, the equivalents are: #include <multimedia/ulaw2linear.h> unsigned char audio_s2u(short) /* linear to mu-law */ short audio_u2s(unsigned char) /* mu-law to linear */ Note that sound files with data format SND_FORMAT_MULAW_8, sampling rate 8012 and channel count 1 are binary compatible and fully interchangeable between the NeXT and the Sun! Note also that you shouldn't attempt to use the full dynamic range of mu-law; "normal" tends to be 9-12 dB down. -=EPS=- 8-bit mu-Law to 16-bit Linear 0 -32124 | 52 -3388 | 104 -244 | 156 9852 | 208 876 1 -31100 | 53 -3260 | 105 -228 | 157 9340 | 209 844 2 -30076 | 54 -3132 | 106 -212 | 158 8828 | 210 812 3 -29052 | 55 -3004 | 107 -196 | 159 8316 | 211 780 4 -28028 | 56 -2876 | 108 -180 | 160 7932 | 212 748 5 -27004 | 57 -2748 | 109 -164 | 161 7676 | 213 716 6 -25980 | 58 -2620 | 110 -148 | 162 7420 | 214 684 7 -24956 | 59 -2492 | 111 -132 | 163 7164 | 215 652 8 -23932 | 60 -2364 | 112 -120 | 164 6908 | 216 620 9 -22908 | 61 -2236 | 113 -112 | 165 6652 | 217 588 10 -21884 | 62 -2108 | 114 -104 | 166 6396 | 218 556 11 -20860 | 63 -1980 | 115 -96 | 167 6140 | 219 524 12 -19836 | 64 -1884 | 116 -88 | 168 5884 | 220 492 13 -18812 | 65 -1820 | 117 -80 | 169 5628 | 221 460 14 -17788 | 66 -1756 | 118 -72 | 170 5372 | 222 428 15 -16764 | 67 -1692 | 119 -64 | 171 5116 | 223 396 16 -15996 | 68 -1628 | 120 -56 | 172 4860 | 224 372 17 -15484 | 69 -1564 | 121 -48 | 173 4604 | 225 356 18 -14972 | 70 -1500 | 122 -40 | 174 4348 | 226 340 19 -14460 | 71 -1436 | 123 -32 | 175 4092 | 227 324 20 -13948 | 72 -1372 | 124 -24 | 176 3900 | 228 308 21 -13436 | 73 -1308 | 125 -16 | 177 3772 | 229 292 22 -12924 | 74 -1244 | 126 -8 | 178 3644 | 230 276 23 -12412 | 75 -1180 | 127 0 | 179 3516 | 231 260 24 -11900 | 76 -1116 | 128 32124 | 180 3388 | 232 244 25 -11388 | 77 -1052 | 129 31100 | 181 3260 | 233 228 26 -10876 | 78 -988 | 130 30076 | 182 3132 | 234 212 27 -10364 | 79 -924 | 131 29052 | 183 3004 | 235 196 28 -9852 | 80 -876 | 132 28028 | 184 2876 | 236 180 29 -9340 | 81 -844 | 133 27004 | 185 2748 | 237 164 30 -8828 | 82 -812 | 134 25980 | 186 2620 | 238 148 31 -8316 | 83 -780 | 135 24956 | 187 2492 | 239 132 32 -7932 | 84 -748 | 136 23932 | 188 2364 | 240 120 33 -7676 | 85 -716 | 137 22908 | 189 2236 | 241 112 34 -7420 | 86 -684 | 138 21884 | 190 2108 | 242 104 35 -7164 | 87 -652 | 139 20860 | 191 1980 | 243 96 36 -6908 | 88 -620 | 140 19836 | 192 1884 | 244 88 37 -6652 | 89 -588 | 141 18812 | 193 1820 | 245 80 38 -6396 | 90 -556 | 142 17788 | 194 1756 | 246 72 39 -6140 | 91 -524 | 143 16764 | 195 1692 | 247 64 40 -5884 | 92 -492 | 144 15996 | 196 1628 | 248 56 41 -5628 | 93 -460 | 145 15484 | 197 1564 | 249 48 42 -5372 | 94 -428 | 146 14972 | 198 1500 | 250 40 43 -5116 | 95 -396 | 147 14460 | 199 1436 | 251 32 44 -4860 | 96 -372 | 148 13948 | 200 1372 | 252 24 45 -4604 | 97 -356 | 149 13436 | 201 1308 | 253 16 46 -4348 | 98 -340 | 150 12924 | 202 1244 | 254 8 47 -4092 | 99 -324 | 151 12412 | 203 1180 | 255 0 48 -3900 | 100 -308 | 152 11900 | 204 1116 | 49 -3772 | 101 -292 | 153 11388 | 205 1052 | 50 -3644 | 102 -276 | 154 10876 | 206 988 | 51 -3516 | 103 -260 | 155 10364 | 207 924 |
ddj@zardoz.club.cc.cmu.edu (Doug DeJulio) (02/26/91)
Does anybody have a utility to extract the data from a soundfile and put it into an ASCII form (ie. digit digit digit newline)? I'm looking to generate a list of numbers that I can feed into Mathematica, so I can do graphs and fourier transforms and other sorts of data manipulation on sounds (I'm practicing 'cause I need to do the same sorts of things on NMR data in the near future). -- Doug DeJulio ddj@zardoz.club.cc.cmu.edu (NeXT mail) dd26+@andrew.cmu.edu (AMS/ATK mail)