blob: 542bd01242a77b7eefd8743c5820648c9bbc4123
1 | /* |
2 | * audio conversion |
3 | * Copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> |
4 | * |
5 | * This file is part of FFmpeg. |
6 | * |
7 | * FFmpeg is free software; you can redistribute it and/or |
8 | * modify it under the terms of the GNU Lesser General Public |
9 | * License as published by the Free Software Foundation; either |
10 | * version 2.1 of the License, or (at your option) any later version. |
11 | * |
12 | * FFmpeg is distributed in the hope that it will be useful, |
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
15 | * Lesser General Public License for more details. |
16 | * |
17 | * You should have received a copy of the GNU Lesser General Public |
18 | * License along with FFmpeg; if not, write to the Free Software |
19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
20 | */ |
21 | |
22 | /** |
23 | * @file |
24 | * audio conversion |
25 | * @author Michael Niedermayer <michaelni@gmx.at> |
26 | */ |
27 | |
28 | #include "libavutil/avstring.h" |
29 | #include "libavutil/avassert.h" |
30 | #include "libavutil/libm.h" |
31 | #include "libavutil/samplefmt.h" |
32 | #include "audioconvert.h" |
33 | |
34 | |
35 | #define CONV_FUNC_NAME(dst_fmt, src_fmt) conv_ ## src_fmt ## _to_ ## dst_fmt |
36 | |
37 | //FIXME rounding ? |
38 | #define CONV_FUNC(ofmt, otype, ifmt, expr)\ |
39 | static void CONV_FUNC_NAME(ofmt, ifmt)(uint8_t *po, const uint8_t *pi, int is, int os, uint8_t *end)\ |
40 | {\ |
41 | uint8_t *end2 = end - 3*os;\ |
42 | while(po < end2){\ |
43 | *(otype*)po = expr; pi += is; po += os;\ |
44 | *(otype*)po = expr; pi += is; po += os;\ |
45 | *(otype*)po = expr; pi += is; po += os;\ |
46 | *(otype*)po = expr; pi += is; po += os;\ |
47 | }\ |
48 | while(po < end){\ |
49 | *(otype*)po = expr; pi += is; po += os;\ |
50 | }\ |
51 | } |
52 | |
53 | //FIXME put things below under ifdefs so we do not waste space for cases no codec will need |
54 | CONV_FUNC(AV_SAMPLE_FMT_U8 , uint8_t, AV_SAMPLE_FMT_U8 , *(const uint8_t*)pi) |
55 | CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_U8 , (*(const uint8_t*)pi - 0x80U)<<8) |
56 | CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_U8 , (*(const uint8_t*)pi - 0x80U)<<24) |
57 | CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8 , (uint64_t)((*(const uint8_t*)pi - 0x80U))<<56) |
58 | CONV_FUNC(AV_SAMPLE_FMT_FLT, float , AV_SAMPLE_FMT_U8 , (*(const uint8_t*)pi - 0x80)*(1.0f/ (1<<7))) |
59 | CONV_FUNC(AV_SAMPLE_FMT_DBL, double , AV_SAMPLE_FMT_U8 , (*(const uint8_t*)pi - 0x80)*(1.0 / (1<<7))) |
60 | CONV_FUNC(AV_SAMPLE_FMT_U8 , uint8_t, AV_SAMPLE_FMT_S16, (*(const int16_t*)pi>>8) + 0x80) |
61 | CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_S16, *(const int16_t*)pi) |
62 | CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_S16, *(const int16_t*)pi<<16) |
63 | CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16, (uint64_t)(*(const int16_t*)pi)<<48) |
64 | CONV_FUNC(AV_SAMPLE_FMT_FLT, float , AV_SAMPLE_FMT_S16, *(const int16_t*)pi*(1.0f/ (1<<15))) |
65 | CONV_FUNC(AV_SAMPLE_FMT_DBL, double , AV_SAMPLE_FMT_S16, *(const int16_t*)pi*(1.0 / (1<<15))) |
66 | CONV_FUNC(AV_SAMPLE_FMT_U8 , uint8_t, AV_SAMPLE_FMT_S32, (*(const int32_t*)pi>>24) + 0x80) |
67 | CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_S32, *(const int32_t*)pi>>16) |
68 | CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_S32, *(const int32_t*)pi) |
69 | CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32, (uint64_t)(*(const int32_t*)pi)<<32) |
70 | CONV_FUNC(AV_SAMPLE_FMT_FLT, float , AV_SAMPLE_FMT_S32, *(const int32_t*)pi*(1.0f/ (1U<<31))) |
71 | CONV_FUNC(AV_SAMPLE_FMT_DBL, double , AV_SAMPLE_FMT_S32, *(const int32_t*)pi*(1.0 / (1U<<31))) |
72 | CONV_FUNC(AV_SAMPLE_FMT_U8 , uint8_t, AV_SAMPLE_FMT_S64, (*(const int64_t*)pi>>56) + 0x80) |
73 | CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_S64, *(const int64_t*)pi>>48) |
74 | CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_S64, *(const int64_t*)pi>>32) |
75 | CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S64, *(const int64_t*)pi) |
76 | CONV_FUNC(AV_SAMPLE_FMT_FLT, float , AV_SAMPLE_FMT_S64, *(const int64_t*)pi*(1.0f/ (INT64_C(1)<<63))) |
77 | CONV_FUNC(AV_SAMPLE_FMT_DBL, double , AV_SAMPLE_FMT_S64, *(const int64_t*)pi*(1.0 / (INT64_C(1)<<63))) |
78 | CONV_FUNC(AV_SAMPLE_FMT_U8 , uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8( lrintf(*(const float*)pi * (1<<7)) + 0x80)) |
79 | CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16( lrintf(*(const float*)pi * (1<<15)))) |
80 | CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(const float*)pi * (1U<<31)))) |
81 | CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(const float*)pi * (INT64_C(1)<<63))) |
82 | CONV_FUNC(AV_SAMPLE_FMT_FLT, float , AV_SAMPLE_FMT_FLT, *(const float*)pi) |
83 | CONV_FUNC(AV_SAMPLE_FMT_DBL, double , AV_SAMPLE_FMT_FLT, *(const float*)pi) |
84 | CONV_FUNC(AV_SAMPLE_FMT_U8 , uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8( lrint(*(const double*)pi * (1<<7)) + 0x80)) |
85 | CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16( lrint(*(const double*)pi * (1<<15)))) |
86 | CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(const double*)pi * (1U<<31)))) |
87 | CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(const double*)pi * (INT64_C(1)<<63))) |
88 | CONV_FUNC(AV_SAMPLE_FMT_FLT, float , AV_SAMPLE_FMT_DBL, *(const double*)pi) |
89 | CONV_FUNC(AV_SAMPLE_FMT_DBL, double , AV_SAMPLE_FMT_DBL, *(const double*)pi) |
90 | |
91 | #define FMT_PAIR_FUNC(out, in) [(out) + AV_SAMPLE_FMT_NB*(in)] = CONV_FUNC_NAME(out, in) |
92 | |
93 | static conv_func_type * const fmt_pair_to_conv_functions[AV_SAMPLE_FMT_NB*AV_SAMPLE_FMT_NB] = { |
94 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8 , AV_SAMPLE_FMT_U8 ), |
95 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_U8 ), |
96 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_U8 ), |
97 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_U8 ), |
98 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_U8 ), |
99 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_U8 ), |
100 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8 , AV_SAMPLE_FMT_S16), |
101 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16), |
102 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S16), |
103 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16), |
104 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S16), |
105 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S16), |
106 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8 , AV_SAMPLE_FMT_S32), |
107 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S32), |
108 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32), |
109 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S32), |
110 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S32), |
111 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S32), |
112 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8 , AV_SAMPLE_FMT_FLT), |
113 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT), |
114 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT), |
115 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLT), |
116 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_FLT), |
117 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_FLT), |
118 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8 , AV_SAMPLE_FMT_DBL), |
119 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_DBL), |
120 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_DBL), |
121 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_DBL), |
122 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBL), |
123 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_DBL), |
124 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8 , AV_SAMPLE_FMT_S64), |
125 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S64), |
126 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S64), |
127 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S64), |
128 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S64), |
129 | FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64), |
130 | }; |
131 | |
132 | static void cpy1(uint8_t **dst, const uint8_t **src, int len){ |
133 | memcpy(*dst, *src, len); |
134 | } |
135 | static void cpy2(uint8_t **dst, const uint8_t **src, int len){ |
136 | memcpy(*dst, *src, 2*len); |
137 | } |
138 | static void cpy4(uint8_t **dst, const uint8_t **src, int len){ |
139 | memcpy(*dst, *src, 4*len); |
140 | } |
141 | static void cpy8(uint8_t **dst, const uint8_t **src, int len){ |
142 | memcpy(*dst, *src, 8*len); |
143 | } |
144 | |
145 | AudioConvert *swri_audio_convert_alloc(enum AVSampleFormat out_fmt, |
146 | enum AVSampleFormat in_fmt, |
147 | int channels, const int *ch_map, |
148 | int flags) |
149 | { |
150 | AudioConvert *ctx; |
151 | conv_func_type *f = fmt_pair_to_conv_functions[av_get_packed_sample_fmt(out_fmt) + AV_SAMPLE_FMT_NB*av_get_packed_sample_fmt(in_fmt)]; |
152 | |
153 | if (!f) |
154 | return NULL; |
155 | ctx = av_mallocz(sizeof(*ctx)); |
156 | if (!ctx) |
157 | return NULL; |
158 | |
159 | if(channels == 1){ |
160 | in_fmt = av_get_planar_sample_fmt( in_fmt); |
161 | out_fmt = av_get_planar_sample_fmt(out_fmt); |
162 | } |
163 | |
164 | ctx->channels = channels; |
165 | ctx->conv_f = f; |
166 | ctx->ch_map = ch_map; |
167 | if (in_fmt == AV_SAMPLE_FMT_U8 || in_fmt == AV_SAMPLE_FMT_U8P) |
168 | memset(ctx->silence, 0x80, sizeof(ctx->silence)); |
169 | |
170 | if(out_fmt == in_fmt && !ch_map) { |
171 | switch(av_get_bytes_per_sample(in_fmt)){ |
172 | case 1:ctx->simd_f = cpy1; break; |
173 | case 2:ctx->simd_f = cpy2; break; |
174 | case 4:ctx->simd_f = cpy4; break; |
175 | case 8:ctx->simd_f = cpy8; break; |
176 | } |
177 | } |
178 | |
179 | if(HAVE_YASM && HAVE_MMX) swri_audio_convert_init_x86(ctx, out_fmt, in_fmt, channels); |
180 | if(ARCH_ARM) swri_audio_convert_init_arm(ctx, out_fmt, in_fmt, channels); |
181 | if(ARCH_AARCH64) swri_audio_convert_init_aarch64(ctx, out_fmt, in_fmt, channels); |
182 | |
183 | return ctx; |
184 | } |
185 | |
186 | void swri_audio_convert_free(AudioConvert **ctx) |
187 | { |
188 | av_freep(ctx); |
189 | } |
190 | |
191 | int swri_audio_convert(AudioConvert *ctx, AudioData *out, AudioData *in, int len) |
192 | { |
193 | int ch; |
194 | int off=0; |
195 | const int os= (out->planar ? 1 :out->ch_count) *out->bps; |
196 | unsigned misaligned = 0; |
197 | |
198 | av_assert0(ctx->channels == out->ch_count); |
199 | |
200 | if (ctx->in_simd_align_mask) { |
201 | int planes = in->planar ? in->ch_count : 1; |
202 | unsigned m = 0; |
203 | for (ch = 0; ch < planes; ch++) |
204 | m |= (intptr_t)in->ch[ch]; |
205 | misaligned |= m & ctx->in_simd_align_mask; |
206 | } |
207 | if (ctx->out_simd_align_mask) { |
208 | int planes = out->planar ? out->ch_count : 1; |
209 | unsigned m = 0; |
210 | for (ch = 0; ch < planes; ch++) |
211 | m |= (intptr_t)out->ch[ch]; |
212 | misaligned |= m & ctx->out_simd_align_mask; |
213 | } |
214 | |
215 | //FIXME optimize common cases |
216 | |
217 | if(ctx->simd_f && !ctx->ch_map && !misaligned){ |
218 | off = len&~15; |
219 | av_assert1(off>=0); |
220 | av_assert1(off<=len); |
221 | av_assert2(ctx->channels == SWR_CH_MAX || !in->ch[ctx->channels]); |
222 | if(off>0){ |
223 | if(out->planar == in->planar){ |
224 | int planes = out->planar ? out->ch_count : 1; |
225 | for(ch=0; ch<planes; ch++){ |
226 | ctx->simd_f(out->ch+ch, (const uint8_t **)in->ch+ch, off * (out->planar ? 1 :out->ch_count)); |
227 | } |
228 | }else{ |
229 | ctx->simd_f(out->ch, (const uint8_t **)in->ch, off); |
230 | } |
231 | } |
232 | if(off == len) |
233 | return 0; |
234 | } |
235 | |
236 | for(ch=0; ch<ctx->channels; ch++){ |
237 | const int ich= ctx->ch_map ? ctx->ch_map[ch] : ch; |
238 | const int is= ich < 0 ? 0 : (in->planar ? 1 : in->ch_count) * in->bps; |
239 | const uint8_t *pi= ich < 0 ? ctx->silence : in->ch[ich]; |
240 | uint8_t *po= out->ch[ch]; |
241 | uint8_t *end= po + os*len; |
242 | if(!po) |
243 | continue; |
244 | ctx->conv_f(po+off*os, pi+off*is, is, os, end); |
245 | } |
246 | return 0; |
247 | } |
248 |